Meet Gor - Tag: database

Techstructive Weekly #78

Sat, 24 Jan 2026 00:00:00 UTC

Week #78

Some weeks are no for hoarding knowledge, I realize this after writing this edition. And this week is that one, the one that you need to put breaks and not reflect that’s done, but see the road ahead, not how far we have come, but where to go next. What to do next, why build something, why spent 2 years on something. Question, think, understand and let it settle. Everyone is doing that. Some aren’t privileged as I am, I am grateful for that.

I am going into the long 3 day weekend with a lot of projects to build, would be disappointing to not have at least one hack from the shed on Tuesday.

Quote of the week

“I carve things of wood because things made by effort are more real than things made by wishing.”
― Katherine Arden, The Girl in the Tower

Oh yes! This hits. I read the book last week and completed the trilogy yesterday. The quote hits home. Why? Just compare the hand crafted, effort rich written code with AI generated code. I guess I don’t have to speak anything more. The rest is on your own right? You can express you feelings about one or the other. It just is similar, but maybe incorrect. Code is a means to an end, just like some tools in our lives are. If I am a writer, laptop or a book is a means to an end, I can’t obsess one over other. I would use either whichever is at my disposal.

Effort and wishing would come in to play when you are doing it for your or others pleasure and to express something which wishing or words couldn’t.

Read

The recurring dream of replacing developers
1. what a flashback, everything makes sense.
2. Programming is not mechanical
3. yet people tried hard to make solve for it as it was mechanical
4. COBOL made syntax readable. CASE tools eliminated typing. Visual tools eliminated syntax. AI can now generate entire functions from descriptions
5. Each advancement addressed a real friction point. Yet the fundamental challenge persists because it’s not mechanical. It’s intellectual. Software development is thinking made tangible
6. Just sheer facts those two. We need to find a way around using AI not away from it.
The joy of being a competent beginner
1. This is really well put. Very relatable. We all have started with some quick competence at something in the beginning and then ignored or abandon after some familiarity of it.
2. This exactly lists why we do that, and the reason is that going beyond that beginner competence is a steep learning curve, initially you are fast but then quickly hit a wall.
3. Most of them give up, the ones that stick, are the ones that somewhat develop a mastery or sort of craftsmanship in the art.
Welcome to Gas Town
1. Agents as code generation orchestrators. This is quite a ambitious thing. I haven’t read the full post. But I can see where it goes.
2. I haven’t either been in situations with 10s of these agents ripping in the background. There are reasons for them and some of them I am trying to overcome.
  1. I don’t have that many ideas honestly, this is flawed in my opinion, my biggest weakness maybe.
  2. I don’t have claude code as I don’t have much to spend on AI. I am limited by free options and some work related subscriptions. I am bogged down by the clumsy free models.
  3. The habit of abandoning a project after the setup is too evasive now. The earlier habit has just got more notorious with AI.
3. That all said, I love this idea, the next step in AI is agents orchestration. Maybe I am behind it, but I had a few ideas, not exactly this but some level of parallel agents running. Not orchestrating. Maybe that idea is speaking more than ever. I thought someone solved it, but nope. Need to roll up the sleeves.
The bet on juniors just got better
1. This is fair. The bet on junior aka me was that I will take the ownership. And this previous year I did. They might have gotten the payback but not quite like AI. The thing that AI might miss is reliability. Not availability.
2. If something goes wrong, I can wake up and roll in. But if some non-informed developer or AI does it, there it could get into a different rabbit hole.
3. I think the more quickly you can show your eagerness to solve problem, actual user problems the better the bet payoff would be. Its not rocket science but is easier said than done.
The challenges of soft delete
1. Nice read. I had experienced it in my first internship. This problem of dead objects. Especially if you are using Django and Postgres. It looked easy to add a field of soft deletion. But the resulting queries could create bottlenecks.
2. Since then I haven’t quite gotten the chance to explore this, this article showed me the different ways to implement the soft deletion.
The toil of blog art
1. Yes, the art of expressing some concept is valuable. AI might have eased it, but true human-eque art is impossible to replicate. The chef’s kiss is what the author is trying to meld in the post about.
Writing first tooling second
1. True, gold. This is to the point and another way of saying, “Show, don’t tell”, so “Write, don’t setup”
2. The blog is merely one possible organising principle, not a requirement.
3. I started the same way. I picked up hashnode, then moved to jekyll and github pages, then used python via markata (waylon walker’s ssg), an ssg that someone else wrote and I loved it, it was what I needed, the control of what goes in and out and also it was easy to see what was happening so that I can change and remove what I wanted.
4. I finally now am rolling my own SSG in Golang and a CMS system. I am not very consistent in sticking to one cms, I have built 3-4 versions of them. But the thing is I still post consistently. Not long form posts, but these reflective posts and short bursts of thinking.
5. Write your brain out first, then eventually it will outgrow to your needs, the system will be formed not shoved in. Like earlier I just used to write long form how-to-guides or tutorials, then I started to write reflections weekly, then link posts tils and suddenly I had 10 types of posts. Article, Tutorials, TILs, Thoughts, Link-blog, Newsletter, Notes, and what not.
Why talking to LLMs have improved my thinking
1. I like the take here. The one aspect of it only.

Writing has always done this for me. What is different is the speed

- That sentence just changed my perspective on LLMs. I was skeptical but now, since they have gotten the powers of thinking, tool calling, I think they are good at talking out ideas and forming maps of different features.
Nested code fences in Markdown
1. This is clever. Never knew this.
2. I think this clears the rule of when to escape the backticks and the fenced code block within one. Really nice to know this. Helps in writing as well as developing a SSG.

Watched

Amp Inc. Raising Agents: Episode 9
- Writing code by hand is over?
  - There will be things where you will have to write the code, but like assembly
  - it can just do things, like give me a cake
    - It doesn’t need the recipe, or hand holding of each task, it can just do it
    - Taste it even and then check if its burned or not, it has a taste or evaluation thing as well
    - It can think about things
  - You need to make the codebase ready for agentic ready
    - It needs harness for testing, good documentation, edge cases, actual problem it solves
Claude Cowork: AGI is here, hheh?
- I loved the video. It showed the flaws and the possibilities of this tool. I think its a step in the AGI, but good or bad, the people will decide.
- The edit button on twitter, that had me rolling out loud. It was a human-esque reply though. “I can see the edit button therefore I am logged in as ABC person” True. Good thinking Claude. Hope you continue in a limited set of thinking.
Claude Cowork
- Again, I was tempted to learn how different people perceive this tool. I never watched her videos. But this video came to me at random and I thought of watching it, it was fun.
- It also showed a good starting point and a legit use cases for people to curse themselves a little less with such tools.
- Developers can do it with writing scripts but laymen can’t oooohhh. This tool should just do that.
We need to talk about Ralp loops
- It kind of is hillarious of how this works! I am not able to wrap my head around it. Like why and how
- What kind of ... Writing it immediately hit me. I do the same thing that Ralph loop does with AMP code free tier.
- The context size is limited, so I have to be wary of the limit, I keep the summary of the thread once the limit is reached and continue a new thread. Wow. Writing actually makes things visible and find the hidden patterns. Gold!
- Watching this video now makes sense. It just a loop for agent to start from where it left off without bloating the context. Superb.
Its time to change your database - from Supabase to Convex

- Oh! The convex database now makes me in awe. It never clicked and all of a sudden it rings bells and whistles. The schema changes the database, that is wild.
- I can see myself using it for my favorite language. GO!
- Its statically typed, so I can catch bugs before hitting them on the database.

Learnt

We are in a time of both awe and suspicion
- I am in awe that I have a tool that can do a lot of things, on the other hand I am kind of having existential crisis on what my job might ask me?
- Like I don’t have anything to add in learning section, I am just figuring things out to learn. I guess some weeks you can’t force yourself to learn technical things, its ok and even necessary to let go of not hoarding knowledge or information (like I learn how to do x in y, etc)
- Some weeks are for converting the knowledge and connecting them to form insights and developing a intuition for the ultimate wisdom. This is that week for me it seems.
- I honestly say, I didn’t read a lot of code, atleast not something out of which I knew. I fixed bugs, yes, wrote code, maybe, but generated code, hell yes.

Tech News

ChatGPT users to have targeted Ads
- Its happening. The inevitable.
Anthropic release Claude’s New Constitution
- I want to read it, will surely read it. But I am tempted to watch theos video first. Will detail read it over the weekend. Looks something is spicy brewing in the AI mind.
Winapp from Windows for making windows application
- Kiind of wild and wired. Who makes them like that? I thought it was like a Microsoft Copilot slop for AI Agent. Thank gosh it was not that.

For more news, follow the Hackernewsletter (#779th edition), and for software development/coding articles, join daily.dev.

That’s it from this long slow slog week. The weekend in India is long. So will be taking some time to invest in learning and building some slop from LLMs. Maybe will build a better intuition for working with LLMs, I know its not a one day process, but I have spent some months in it and continue to do so. See you next week!

Happy Coding :)

Techstructive Weekly #76

Sat, 10 Jan 2026 00:00:00 UTC

Week #76

It was a good start to the year, finally doing something that I had struggled to do for the past year or so. AI Assisted Programming. Yeah! That was something I finally somewhat understand, and can do it without feeling a slightest of grudge or emotional drama. It took a while to realize it, but here we are. 2026!

I don’t know how I feel right now. Its quite a good times to be in tech.

Quote of the week

“If you want to know what a man’s like, take a good look at how he treats his inferiors, not his equals.”
— Sirius Black
― J.K. Rowling, Harry Potter and the Goblet of Fire

This is true. I think we should treat people with kindness irrespective of their position. Position is no match for one’s love and care for us. They might be doing with their own purpose and needs but they still show it.

Created

World Atlas Game
- This is vibe coded in a day. Gemini CLI and Amp. Just ripped it
- I read Golang was good to work with AI Agents, thought of building some backend with it and lo behold, it did one shot it almost. For the frontend I choose Vue. Surprisingly its a great UI. I am honestly impressed. I didn’t write a single line of code, let even see.
- I always wanted to make this, but was very lazy to do all of the meddling with the boilerplatey code, it just did in a few minutes.
CMS with ssg
- Yes, this was something I have built twice or thrice, it was another shot. I wanted a blog that can just save to the sqlite db and fetch aas cronjob every 6 hours to build the site with ssg. I just gave it and it did. Its not great, it has still qwirks, but making it better over time.

Read

Dialogue between a developer and a kid
1. This is hilariously funny.
2. What a real developer is? Who knows languages? No, who knows how to code, No! A developer is someone who sticks to a problem when everyone has given up.
3. This conversation feels like me and my friend. My friend is the reason I am here today. He knew programming well. I was inspired from him, he gave me advice to learn one programming language, I was boasting about python, C and C++. I feel like a kid here. That was 7 years ago, time flies by.
Fear is not advocacy
1. This is real advice. People are hyping about the next workflow to 100x our productivity. Its ok to be 1x and still push less bugs than 100x and push 1000 bugs.
Quick and dirty print debuggin in Go
1. This is cool, we make logging a mess. For logs we need to have separate scripts to get relevant data. How much chaos it can be.
AI Did Not Take Your Agency. You Handed It Over
1. True. LLMs amplify ambguity.
2. If LLMs don’t have agency, they don’t choose constraints. Well put.
On not using Django
1. I don’t quite get it. Maybe its true. Django provided a good start but then it was like a lock in.
2. With LLMs its quite easy to generate the boilerplatey code that django provides out of the box, so that demand is lost?
3. Its not the only reason django is here right? It has extensions, best python community and even more best documentation.
4. I think it will be the best framework to build with LLMs in the future if the ecosystem continues to improve
6’7’‘ is not Random
1. This is so true

In the 1990s, a “middle-class job” was enough to buy a house. Being “6 feet” was enough to be tall.
In the 2020s, the middle has been hollowed out.
To be “wealthy” now requires a crypto-exit or a tech IPO (The Economic 6’7”).
To be “famous” requires global virality (The Social 6’7”).
To be “attractive” requires filters and surgery (The Aesthetic 6’7”).

AI should be free software
1. Yikes, this looks like a good take on LLMs being free and open weight.
2. If not, the larger AI labs might offer ads into the LLM suggestions. This, just the thought of it makes me wiggle with fear. It might push us in wired directions.
3. The point of drawing a line of “our goal” vs “model’s goal” becomes hazy and it just doesn’t align with human values.
4. Its a pretty hard problem to solve if it goes in a bad direction, which it seems to be at the moment.

I just completed reading Harry Potter #4 the Goblet of Fire. It was amazing. A good start to 2026 in reading. Hoping to complete the series in February.

Watched

Designing Data Intensive Applications: Chapter 1 and 2
- It was a great overview of the database systems. I like how he explains the p50, p90, and all metrics. It makes sense without getting into too much of details
- Also the diagram of the OLAP and OLTP databases and how it fits. It made sense.

AI codes better than me, now what?
- This is really changing. It can write code, better than me. That’s when I started to use it as a partner that knows a lot of things but gets overwhelmed and like a junior does a lot of things.
- Guiding it, reviewing it, and also understanding myself what it actually does is co critical.

Database Internals:Chapter 1
- The difference of the OLAP and OLTP database is so nice.
- Also the differnce of column based vs row based database type is clear from this. Makes sense and intuitive as well
- The Binary tree also makes sense.

The year I stopped writing code
- This is interesting and eye opening. It actually gave me the reason to be active while working with LLMs.
- Reviewing is hard, most developers avoid it, that’s the part you need to be doing, in order to be a better one. That point I had ignored and it has came to haunt me in the year throughout. This new year though, will be different. I have decided to take LLM generated code with a grain of salt.

Learnt

How to ship code with Cursor
1. I explained my euphoric moments in the week where I discovered the debug and ask mode in Cursor. It helps me to understand the problem, learn something. Which agent modes doesn’t let me.
2. I can pause and let it show me what is happening, I can read and share with it, what I think and have a conversation and not just make change all the time. The switching mode was liberating, I think these models should know when to ask and when to execute.
Read csv from Pandas in python needs quoted string for multiple commas
1. If you have n headers and have n+m commas in the row, pandas’ read csv function will break
2. Because there is ambiguity in which comma is the header separator and which is the actual text comma.
3. Use quoted string for the text if it contains comma.

Tech News

For more news, follow the Hackernewsletter (#777th edition), and for software development/coding articles, join daily.dev.

That’s it, a good start to the year, looking forward to a good and great year 2026, will it be slow? Probably not, but the year seemed to start slow. Looks good for now, we already have a lot of things already to unpack from the last year advancement. 2025 was pivotal for anyone in tech, 2026 onwards it looks like a year to build and carry that momentum.

Leave a comment

Share Techstructive Weekly

Thanks for reading Techstructive Weekly! Subscribe for free to receive new posts and support my work.

Techstructive Weekly #75

Sat, 03 Jan 2026 00:00:00 UTC

Week #75

Oh the middle of end of 2025, and the beginning of 2026. This is a wired post. I let myself back, reflected more. I know it was a fast paced year for software development. People are here to prove it. I just laid it out straight that its never been valuable to be a human, a distinct, natural and earnest human.

I must say I completed a non-technical goal of 2025, I have my novel’s first draft in my hands(in my google drive). Its very rough, I want to revise it, I want to remove the fluff, but it is there in its entirety. 33 Chapters, 85K words. 45 days of writing span across 6 months, I did it. I am pumped to revise and write the next novel. The hunger to write has never been higher for me.

Quote of the week

“Grit is never bad, grit with wrong intent, for wrong purpose is definitely bad”

I had a grit to write a story, I will write it. It was a grit, not a goal. I had goals for it for the past 4 years, yet when I made it a grit, here it is, I am on top of it. If a grit was something to think bad of someone, or look down on someone with grudge, it is a bad thing. A wrong grit is something born out of desires, rage, frustration, anger or even jealously, None of it could be bad per see. But if the intention of those desires, rage, jealously is for ill of someone then it could be bad, not if it is born to uplift yourself from the ground, after having thumped by life.

Sometimes, grit is something life give us, throws at you, you need to learn to handle it. It can’t be thought, it needs to experienced. You have to fail, you have to struggle. But here for me, the struggle, the resistance to write finally bent its knee against my grit (not boasting, nor arrogant). Just fierce grit born out of the desire to be a better person, to give back to a human, to say a kind thank you.

Wrote

2025 Year Review
- Feels good to write this posts. I never miss them! I write less post this year. Just 2 or 3, but I have written a lot of SQL like learning log posts (53 of those).
- Not to forger I am also writing my thoughts on the things I learn and read here. So 52 articles for each week. Easily making up 100 writing pieces.

I completed Advent of SQL with 15 posts too, here are the remaining, which I completed on the weekend. I learnt a lot, it was a good one, the problems ramping up gradually, then the lore for each post was so good.

Read

The future of software development is software developers

technical practices that can dramatically shrink delivery lead times while improving reliability and reducing the cost of change, with or without “AI”

- A banger of a post. This is expressing that AI is just a shift in a toolset, or maybe even a abstraction of the language. We will still have ambiguity that a human needs to understand in order to deliver a software.
- We can see from his experience, developers were written obsolete from time to time, and each time it was different, more potent than the other, but here we are.

On top of all that, “AI” coding assistants are really nothing like the compilers and code generators of previous cycles. The exact same prompt is very unlikely to produce the exact same computer program. And the code that gets generated is pretty much guaranteed to have issues that a real programmer will need to be able to recognise and address

- This is true too, we are just automating and generating crap faster, code is always crap until distilled and refine with each iteration to the needs. We just now have a better or worse iteration cycle, a machine that can spit out code like tirelessly, we need to vet and test it.
2025, The year in LLMs: Simon Willison Weblog
- Boy! That is a lot! I have been saying “overwhelming” word was not sufficient to describe this tend of LLMs in 2025. This explains the reason
- We had LLAMA falling, Gemini gripping, OpenAI still on the top yet cornered neck to neck suddenly with Chinese Labs and Anthropic in its own league. We saw the sudden rise and sudden dip in vibe coding, people thought “We can be programmers! We don’t need developers anymore, hehe” to “Damm! Do I need a developer to debug this?”. That was a funny thing to watch (as a developer)
- The images and 6 second video clip generated by AI are mind boggling, we saw from Sora and Nano Banana what havoc they can wreck if put in untamed hands.
- Local models are getting good, but the speed of the cloud and advancement over the other side is rocketing. There is also this trend of cli based agents. Claude code just set the trend and let 100s of cli agents rip off in the months to follow. Those are still released by new companies every now and then.
- Slop, yeah! We had less human slop than we needed AI right?
- Thanks to Simon Sir for this awesome blog. It finally gives me a relief to read so many thing have happened at a glance
Andrej Karapathy’s 2025 LLM Year in Review
- This was more of a reflection post, of how his mental model has changed and how things are building up. I like it. It was a interesting and highly technical perspective.
- His opinion of LLMs as Ghost is so liberating, as it actually threatens me from my identity if we compare it with humans. Ghost makes sense, even dismissive it as a slave sort of relation right? Not in a bad way but kind of inferior relation for LLMs with humans.
- Agent that lives in the terminal is practical, for a developer or a human who understands what they are doing, they know what they want, its just too much menial for them to spend the energy on. I agree.
- There is a lot of work to be done, developers, don’t strap your belts, hone your hammers, its going to be needed.
- Also his post:

Andrej Karpathy@karpathy

I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become

5:36 PM · Dec 26, 2025 · 15.9M Views

2.58K Replies · 7.24K Reposts · 54.5K Likes

- Could this have been more accurate! Right note to end the year.
- Vibe coding last year, now this is the trend we are surfing on, this will last decades.
Local LLMs are how nerds justify a big computer they don’t need
- Curiosity gets the better of them. I have a 8GB device, I can barely run a 1B parameter model. I get frustrated but have nothing to complain. I can use ChatGPT in temporory mode, or incognito mode if I don’t want it to attach it to the memory. I don’t see using local models on scale is justifiable just yet.
You are not dumb, you just lack the pre-requisites
- Yeah! I have started to learn SQLite and since 2 years made a Brilliant org streak. I feel good taking on advanced concepts soon.
- Basic and a good foundation helps you pivot and branch off to wide possibilities.
13 Tips for Writing a technical book
- A handy little thing to remind myself, this is inevitable for me. I would write one. Not this year probably. But I would surely write one, my gut, my instinct is not false on this.
- I would this then. Great advice for just being curious.
Shipping at inference speed
- This is a good post to say that we have different ways of using LLMs at this point and nothing is permanent. Every month or weeks, this is changing. Adopting a new workflow is like juggling circus art.
- Codex is something I haven’t even touched, Claude code too, never. I have used Amp, Gemini CLI, Warp and Cursor the most.
- I love those, those are cheap or even free, they help me understand what I was about to do wrong. They have never produced anything right 100%. I always needed to understand what was I supposed to do.
- Is this true “>The important decisions these days are language/ecosystem and dependencies” Maybe but I don’t see that. Its kind of true, but not in a big way. The major things are the flow, the edge cases and the intuition for the problem for it to be ale to understand.
- This actually surprised me”> Go wasn’t something I gave even the slightest thought even a few months ago, but eventually I played around and found that agents are really great at writing it, and its simple type system makes linting fast.” I want to try it now. I have ton of go projects.
The internet is a net negative

We’ve maximized information and accidentally drowned wisdom

- Hits home. This is good observation and a perfect critique, not over cynical, nor too loathed. Its just helplessness to avoid the battle of the mind and the heart.
- The business of the world has forced humans itself into a trap. What an irony we live in, creating a cage for ourselves. Besides slaughtering nature into it too.

All that time, that irreplaceable human attention, fed into machines that convert consciousness into quarterly earnings.

- That hurts badly. We are loosing are attention to these machines. We need to get it back. The time, the wisdom and the boring tone to our lives.

The optimist in me is still here. Still hoping.

- This is great piece of writing. I love it. Want to write essays like this.
- Thanks for writing this Kenneth, you have inspired some spark for me.
What I learned writing Gleam, after coming from Python
- Top to down approach. This just shifts from taking the problem and boiling it down to the input and output. Wow! This just made so much sense now.
- We can define the main API as the function that takes something and returns something. In between the intermediate steps, we can then decide what each component of the result will come from.
- I need to try hard on learning functional programming this year.
Engineering is becoming bee-keeping
1. I like this comparison quite a lot. Swarming agents is what its happening. And the realisation that code was the thing that doesn’t matter, the thing that matters is did we solve the problem

Honey shows up at the end. That’s what matters.

And bees can sting. Without the right gear and practices, you get hurt. The protective suit, the smoker, the careful movements. In code, that’s patterns, documentation, tests. The guardrails that keep the stings to a minimum.

Working like this is exciting. There’s a playfulness to it. You can try things without committing. You can explore without sunk costs weighing you down. You can work on three features at once because you’re not holding all the context in your head anymore.

You cannot not lead
- This is so subtle, yet perfect. You lead by good or a bad way.
- You cannot say I was not the leader when you are the only person building and maintaining it. You lead by examples, good or bad.
- Average sucks you know? You are either good or extremely bad. You cannot not lead.
- Wow! This post is so perfect, not only fits the manager but also every human, a elder human trying to teach or lead a younger one. The younger one learns from the examples and behavior of the leader.
Git Integrations is ten years away
- This is hillariously funny. I can’t imagine VS Code team coping up with git integrations in 2025
- I don’t blame them entirely, at least they realize it is missing. With these LLM assisted coding, they decided to ship it finally. We have one instance of AI assisted coding helping VS Code ship faster (after 10 years).
- Learn Git, true. I alway 100% of the times use the cli. NO aliases, no agents, just CLI commands. git add, commit -m, push, pull, merge, rebase whatever. If I don’t know, I google it, read the ai overview and straight to the keyboards.

Watched

CMU Databse System #3 Database Storage: Files, Storage, Tuples
1. This was a good lecture on the different storage hierarchy of the storage. The top there is the pages, the blocks of memory that database fetches for individual records or tuples. Then there is the blocks of memory on the databse file itself, and the actual disk of storage.
CMU Database System #4 Memory Management and Buffer pools
1. Ok OS is not our friend, we need to manage our memory ourselves. This went wild, I thought managing memory was like shooting yourselves on the foot, but not for DBs.
2. So we load the database file, from the disk into memory not as full, but chunks called frames, where each page is contained in the buffer pool. Interesting, this is done in the actual ram or the memory not full at once.
3. So this makes it the different algorithms to decide which frames/pages to keep and evict (remove)
4. There is a difference in lock and latches, a lock is something that protect the database logical content from other transactions i.e. the data to write or avoid corrupted reading
5. However a latch is something that helps in preventing the database internals from other operations, its only for an operation not a query. Its like a mutex.
6. We can’t rely on OS, as OS doesn’t know what are we querying.
7. There are like half a dozen implementation of replacement caches like LRU, Clock, LFU, LRU-K, ARC, etc.
How I parsed billions of rows for every user in 2 seconds
1. Wow! I like these videos. I learnt a lot too. It was passive knowledge true. But I came to know that these things are at least possible.
2. Clickhouse as a database, the queries, we can use Material views which can be used as a CTE almost but on the fly, Endpoints to query them as a URL. WOW!
3. The optimisation was based on the clickhouse features only, not sure if it would have been possible without it.

Learnt

SQL Recursive CTEs
1. We can define a recursive CTE by referencing the CTE within it
2. We have a single row (could be multiple as well) as the base case
3. Then we define the recursive part, by referencing the cte as the table we are fetching the records from with the data queried to it as the parameter.
4. Used it to solve Day 14
FTS in SQLite (Full text search)
1. I learnt how to write a query for FTS and construct like a index for searching across tables.
2. This is efficient from the string comparison as we don’t have to define how to look it up, we just define what we want. The algorithm and the query planner does it efficiently for us without storing it separately on disk.
Golang might just be better than python for writing LLM generated code
1. Its simple has a type system. So it makes it easier for LLMs to generate valid code with correct checks in place.
2. I need to experiment it with to understand the nuance this has.

Tech News

Leave a comment

Share Techstructive Weekly

Thanks for reading Techstructive Weekly! Subscribe for free to receive new posts and support my work.

Well! That was a thumping start to the year 2026. I have bright ideas and a new canvas to paint. Looking forward to have things running and working in my favor over this year. After a slog and slump for 2 years, its time for me for redemption. I can see a hope, hopefully you can too. If not, you will soon.

Happy New Year!

Happy Coding :)

Techstructive Weekly #74

Sat, 27 Dec 2025 00:00:00 UTC

Week #74

It was winding down week, 2025, ending slowly, the biggest irony. The year was full of fast-paced, unprecedented models just dropping here and there. It was hard to keep up with the progress. It was getting overwhelming, no one is used to such a level of information. I was dreading with so much power in our hands. I couldn’t handle it and refrained it and thought it was better to focus on learning new things. I kept using AI tools at work, not by choice, but initially through force but then through necessity.

People made developers go fast, but they realised,

Churning code was never the bottleneck
Solving problems and understanding the business needs was the core goal.

Yet, people thought and here we are with tools like Cursor, Claude Code, and 10 other CLI tools. I learnt the hard way to leverage this tools, to use only when the code was throw away, some scripts, and getting insights from logs. But never on actual bugs and features, it was draining, lacked the joy I get from actually making it.

So, at the end of 2025, I am a decent developer who cannot use AI tools. Yes! give or take, I would love to learn more in 2026.

Quote of the week

“I don’t fear a future with AI. I fear a present without thinking.”

Yes! That is what I believe in this year. I have stopped thinking it seems. If in 2026, we move very fast without thinking, I need to change for the better. The directions we take, everything is a decision, but on what ground, on what thought. If the thought are not 90% yours, you are not thinking enough.

Think, don’t let AI do it.

Created

Advent of SQL 2025 in SQLite

Read

You’re not burning out, you’re essentially starving
- This was a good one
When you truly chase your highest potential, everything you thought was burnout will melt away. Because you weren’t suffering from too much work, you were suffering from too little truly important work. Like a boy who thought he was full until dessert arrives, you’ll suddenly find your hunger return!
Some really good points
Pause once a month to make sure you’re still on the right track. Stop once a year to triple-check you’re on the right track. But never get off this path towards your highest potential. Anything else will starve you existentially
This is true
We’re optimizing for less suffering instead of more meaning.
Yes
I woke up today so excited to get to work thinking it was Monday morning already. Instead of jumping right into it, I spent all morning making breakfast and playing with my kids, then wrote this post. When I’m writing about something personal, 1,000+ words can easily flow for me in an afternoon.
Just read the post!
Don’t become the machine
1. This is well put.
2. I kind of hate this argument.
3. Why are we comparing ourselves to machines in the first place? We can grind, but with thinking what actually we are doing.
4. Because I equate grind to consistency, it sometimes feels like grind, and we need to overcome that emotion of letting it overtake us. But most of the days, the grind is a joy, we do it because we feel like doing it.
How I left youtube
1. Man that was a good read!
2. I resonated with this a lot
  This duality is exhausting. It forces you to lie by omission to people you respect. You can’t tell your team, “I can’t take that ticket because I need to study dynamic programming.” You just have to work faster.
  I respect people above and behind me, but I too needed to move in life, support the things I was responsible for, get out of the grave situation I was pushed into. For that, I took some decisions, which I tried for, but nothing came off it, I wasn’t quite sure about the switch and left it when the offer came. Stranded here. I am feeling good here, but if I am not in another company by the end of 2026, something is wrong with me.
  Let’s see!
3. Good lessons
  - Don’t say: “I tweaked the YouTube watch-time algorithm using X variable.
  ”
  - Do say: “I optimized a high-throughput distributed system to prioritize user retention metrics, reducing latency by 150ms through a custom caching layer.”
  Man!!
  My final conversation with my manager was heart-wrenching. I had prepared a script, anticipating a counter-offer or a guilt trip. Instead, I was met with soft and understanding empathy.
  Yep
  The interview fatigue is real, and the conversations are hard, but the clarity you gain on your own value is worth the struggle.
4. Got to go through it once and then there would be no stop for growth.
  Grass looks green on the other side always! Damm
Text editors should be worse
1. Ok! I agree and disagree. You need to have a zen mode in your editor, which just is bare bones, and one for full fledged stuff like LSP, AI-auto-complete, syntax highlighting and what not.
2. Editor is just a tool, it can’t code on its own(in 2025, still needs prompting), similarly to use it, it needs preferences.
Structured output can create fake confidence
1. Spicy take and true! Somewhat true
2. If your task is complex to get things out from image, or understand the context, it might hinder the quality.
3. But if your task is to simply do something straightforward tool calls, structured output beats everything.
Writing HTML by hand is easier and cheaper then debugging your SSG in 2025
1. Wow! We are moving at a pace where generating html from LLM is getting easier (not cheaper yet!) than generating it by code, whew! What a time to be in.
Guide to Local LLM Models
1. Ok, the VRAM and RAM is somethign is quite critical. If you have less RAM and much VRAM, its no use, you need to have sufficient RAM in order to run a good enough model, VRAM wouldn’t handle it.
13 Years of Rust and the birth of Rue
1. I see this a lot! People creating something that they wanted but didn’t had the mental energy for.
2. I see it as draining rather. I can’t watch it write code for me, its a dreading feeling to be in for larger durations.
Vibe Coding is broring
1. It is pathetic, really. Watching it clog some code and done. Sigh what is left out then, to read code? Who loves it.
2. Vibe coding is cool and good if you just want the product in your hands, but if you care about the craft then please write it.

Watched

Bublesort is useful
1. Yes, this is kind of nuts
2. Buble sort is the lowkey high value thing to learn and know of.
3. VIsualizing any sorting algorithm really makes you understand the flow better and it clicks almost everytime.
The Fundamentals by Kelsey Hightower
1. Ahh! How many people will say this, but yet we can’t follow it
2. Everything boils down to the fundamentals, having the basic thing to understand when something goes wrong. Rather we make it complex in order to be percieved as smart and even oversmart.
Will Turso be the better SQLite? Interview with Glauber Costa
1. This was a great interview. I love the mentality.
2. If software built with community in the Linux community can sustain after almost 3 decades, then why can’t a embedded database like SQLite can?
3. Turso is Linux Community but for SQLite (minus the toxic leadership)
4. Pekka is a great, humble and smart leader to be leading the Turso, SQLite rewrite in Rust.
5. I want to contribute to SQLite, but it feels I don’t know enough everytime I touch it, also I started learning SQL for this. I have gone so far and now there is no way I am turning back.
6. I had one itch for geospatial exploration in SQLite for Mumbai city. This weekend might be the time to do it, maybe next year.

Learnt

Using Tool calling in Google Gemini API
- We can pass the tool as Code Execution block and it can essentially work as an agent in the api. This is a superpower to have.
- Can imagine people creating workflows and all sorts of things with the api in gemini, and gemini, kid you not is a really good model, like it can just do things. (Not complex things, but simple things, it can do really well)
I played with Grok Imagine
- Oh boy! These image and video models are getting really out of hands. I just uploaded my photo and boy came a introduction about me, like a one sentence greeting, but it was scary that it can do that, that quickly.
- I underistimated how quickly these models will evolve, we might plateau out eventually, but still the progress made is mind boggling.

Tech News

Well its Christmas and end of year.

God! Dam! This AI labs have learned something from last year. We don’t have groundbreaking models now! 2025 was a rollercoaster.

For more news, follow the Hackernewsletter (#775th edition), and for software development/coding articles, join daily.dev.

But here’s my wrap in 2025

Wrote 52 articles from #23 all the way to #74
Wrote 50+ articles on SQL on #sqlog
Learnt about SQLite and solved 15 advent of sql #advent-of-sql-2025

All in all around 120 posts on my blog #blog

See you next year!

Merry Christmas and a Happy New Year!

(oh before that, there will be a 2025-yearly-review post)

Happy Coding!

Leave a comment

Share Techstructive Weekly

Thanks for reading Techstructive Weekly! Subscribe for free to receive new posts and support my work.

Techstructive Weekly #73

Sat, 20 Dec 2025 00:00:00 UTC

Week 73

A pretty slow and sluggish week, but some momentum carried in the end. There was a disappointment after a glimmer of excitement when gemini 3 Flash dropped for the experiments that I was running for extraction of documents. It was maybe just the timing, but after this seeing code execution from chats, it was amazing. A good end to the work week.

I have continued to write Advent of SQL for the past 4 days, and brought back the streak of writing SQLog. I was not able to ship some code over the past weekend. But this weekend, I am pumped. I have time sorted out. Would be shipping some improvements in the website. Oh! I actually added snowball and particles based on season on my website (only index and post pages).

Last weekend, I wrote 4k words in a single day in around 3 hours. I was pumped to get all in the coming week, but only managed to get 2k words in the 5 days of the week. Struggling with consistency but trying to get hang of it. I am planning to wipe the story off this weekend, a 3-4 hour session could do it. And another goal of the year completed. I can’t be more happy. A novel, first draft at the end of 2025.

Quote of the week

Revenge may serve a purpose when it corrects injustice, but it loses its meaning when it is used only to satisfy one’s ego

I was continuing to read the Harry potter series, this time the third “The prisoner of Azkaban”. I was observing Harry’s thoughts and his intentions, he wanted to take revenge, hinted by Malfoy, he found that intriguing and wanted to kill Serius Black without a reason. However, Sirius Black wanted to kill Peter with intentions of taking revenge for his betrayal of his friend’s trust. For a few moments, Harry’s intentions were wrong. But time(Lupin) corrected his perspective and he got on the right side. Revenge is a good thing if used with proper intention without attaching your ego and selfish interest. However when rage hits you, you loose the decision to take proper and fair actions. One needs to be capable of handling rage and directing it to a positive outcome, or letting it burn you, because to shine bright, you need to burn.

Read

Deliver Code you have proven to work as a software developer
1. Right on! So true. merging slop without review, even slightly running against a manual test case can give you a good confidence and make you a good engineer.
2. With more code being generated and as easy as a command, it becomes rather intimidating to review code, to accept suggestion and produce more code. But code is not a magic wand its some assumptions crumpled with logical validation, both of them contradict yet when done right, creates a software that people use and breathe.
3. I agree to the manual as well as automated testing and LLMs will follow the pattern. If you already have tests in the codebase, it will make sure the test suite is updated when it makes a new change. There are obvious and unavoidable circumstances when you’d have to check the changes with manual test, its something that comes with the plate in the software engineering role. There is no denying in this, its a fact not an opinion.
4. Software and accountability is opposite side of a coin, you can’t let software account on its own, humans have bought its existence from their imaginations and manifestation, you need to validate and prove the thing you wanted to build.
Duplicate Reports
1. Testing guys is the vibe of AI, testing code is becoming apparent as AI can produce code in matter of seconds. Learning the fundamentals has never been so vital.
The time elemet that should actually do something
1. Another div without a functionality problem. People create standards and forget to adhere. We have so many protocols, people and developers follow them, but there could be places where no one’s actually paid any attention.
2. I am surprised there is no element to depict a time for an search engine to rely on, it relies on external factors like datepublished and other in the schema, wired. Even Google doesn’t care about this tag! Pathetic.
Stop crawling my html, use the API
1. This is so funny, the API is in front of the user.. No LLM, but its so lazy to hit the API.
2. Maybe we need another protocol for how AI should scrap data from websites, but scrapping is a thing that doesn’t have a standard, or rather no one would follow it.
The strange case of engineers who dismiss AI

Programming is a task; software engineering is a role

Maybe it feels threatening to their identity, I mean, your expertise is wrapped up in being someone who can write code. Some tool threatens that? Of course you want to dismiss it.

- Wow! What a statement. Just bangers.
- I like the way of relating things, the difference in AI coding tools from 2022 and 2025 are like Internet Explorer 11 and Chrome. It really is, they are also getting faster and cheaper (maybe not but still).
- I think I bet on it, use it to ship more.

Wrote

Advent of SQL 2025 on databaseschool.com
- I am solving the problems in SQLite, I want to dig deep, learn more ways to solve one problem. Deepen the knowledge of using various constructs.
- So far I learnt about UNION, differences in INNER, LEFT and RIGHT JOINs and proper usage of CASE WHEN THEN END

Watched

Gemini 3 Flash
- Wired model. Gemini vibes. But it does something good. It is fast. Hell fast from GPT 5.2 and what was that 3 Pro, what slow lazy models those are.
Mostly Technical: Hearts and Minds
- Boy I have some thoughts here.
1. Aaron’s AI Stack > Claude Opus 4.5, Amp Code, Code Rabbit for review
2. Ship code, no one cares how its done
3. Have an idea > research > plan > throw it to AI > look at it, stare at it > ship it
4. What is the perfect abstraction, no, what can I get shipped.
5. I like to code, but I loved implementing ideas, now its easier to code with AI, it knows the patterns and abstractions. You have to eyeball the code slop it generates.
6. AI to check in AI, I was too not sure of, but lately the Seer bot from Sentry is so cool, it picks up grave stuff.
7. You need the human, flavour is the juice.
8. People can produce code, but not software, you have to have a point of view
9. I have more then ever to build, the need of software engineers is going to get more. Maybe, I don’t know. They would need a person who can steer them.
10. Its a great time to study systems, and not specific frameworks. Argh
  1. I hate to say that, but I disagree here. We need to know the tools, not specificity but still, humans are nerdy people they can’t live without doing or learning something, even if that is pointless.
  2. Learning Systems, true, I agree wholeheartedly to that. Maybe he means in the terms of content creation. People are not going to watch or read such specific guides to tech framework and tools, but broader skills than technical details.
  3. But I still think having the knowledge of specific tech or tool will give you the edge over the one slopping and producing slop when the time comes.
11. Human’s shared experience is something I am starting to consume more. Or rather consuming just that. No one likes AI slop, look at hackernews, people are reading experiences of x person using y ai tool to get things done.
12. The point of Aaron on shared experience on Pride and Prejudice written by human, is something people are still consuming and talking about after a decade or more. But what about PaLM? Do you remember the model? Noooo. We need human connection.
13. The idea train from Ian is contagious, I am running it something on my brain to think of something to make in SQL or some code.

Learnt

Google AI Studio has Code Execution ability
Difference between INNER, LEFT and RIGHT JOIN
- In INNER JOIN, the rows are include from either of the tables in a single relation only if the condition is met.
- In LEFT JOIN, the records in the first table (left) are included no matter what the condition is, even if there is no relation in the right or next tables, it would populate a NULL record for those.
- In RIGHT JOIN, the records in the last table (right) are include no matter what the condition is, even if there are no relation in the left table, it would populate a NULL record in the left table for making the right record shown in the final result set.
If you want to merge two tables, you need to use UNIONto make sure duplicates are discarded.
- Its like a set, two tables, you need the union of both the sets, and remove the duplicates.
Use count and case based increment when we need separate count from the same table.
- SELECT COUNT(CASE WHEN status = 'approved' THEN 1 END) AS approved_count, COUNT(CASE WHEN status IS NULL THEN 1 END) AS in_review_count FROM hotline_messages;
- Here, we are counting approved and in review count form the same table.

Tech News

Google Releases
- Gemini 3 Flash
  - A good jump, the preview model hmm, from the tests I have done on OCR, it struggled from teh 2.5 Flash. Maybe let’s wait to make it stable.
- Function Gemma
  - A 270 Million parameter fine-tuned model especially for function calling and following instructions
  - This looks really promising, can’t wait to build something on the phone or the cloud with it.
- T5Gemma2
  - This is a 270 M, 1B and 4B parameter with multimodal and mulitlingual capabilities, it has a long context of 128K.
  - This again stands a good balance from lightweight to intelligence ratio, highly a good candidate for making people transition into full on AI systems. I don’t know if it would lead to good or bad outcomes but a good step from Google.
Open AI release GPT Image 1.5
- Competition is harsh here, Nano-banana is so good. yet its a little behind now. I don’t know much about Nano banana Pro, but Image 1.5 might fall behind in its speed. Look at Nano banana, its zip zap image, Imagen is a slog. Not sure from the API though.
Astral launches ty, a fast Python type checker and LSP
- They are single handedly taking Python ecosystem to a spin, they are crushing it.

That’s it from this week, pretty much a good week. Things are shifting in software from writing code to testing it. Not sure what awaits for us, the software engineers.

For more news, follow the Hackernewsletter (#775th edition), and for software development/coding articles, join daily.dev.

See you next week!

Happy Coding

Leave a comment

Share Techstructive Weekly

Thanks for reading Techstructive Weekly! Subscribe for free to receive new posts and support my work.

Techstructive Weekly 68

Sat, 15 Nov 2025 00:00:00 UTC

## Week #68 It was a great week. A consistent clean week. I wrote around 9k words on my project. I feel good again. I didn’t do much on the weekends as usual. I am slowing down. I did a lot of things at work, experiments, and more experiments on VLLMs and parsing documents. It’s fun times. I think for this weekend, I have plans as this post will cover what my mind is fixated on for the moment, Golang and TUIs, and a pain that is daily buzzing me, cleaning log files to get the actual data. LLMs are good at it, but take a bit of time for such trivial things. Apart from this, I have also been reading, a lot of reading, I completed a book of 300 pages in 3 days and found peace. Started one more. I want to create a webpage full of my notes and reviews of all the books I have read so far, not sure if I can do it before the year ends. ### Quote of the week “To understand emotion, don’t seek the outcome, or the reaction, go deeper, find its intentions.” This is my quote, I have observed from reading “Goddess of the River” by Vaishnavi Patel. The character of Karna, and his opinion about friendship with Duryodhan, is so subtle yet this is what I can comprehend. He went for the outcome. Duryodhan made him the king of Angadesh by giving him his part of the kingdom, to make him royal and worthy to compete with Arjun. He didn’t see the intention, his intention was to defeat Arjun. The revenge was in the minds of Duryodhan, not peace or friendship. He made everything in life less important than his loyalty and debt to repay the friendly help. If he knew the intention, he would have been on the right side of the war, maybe the war might have been averted. But no, he wanted to feel recognized and equal to Arjun too. So, look beyond the outcome, the intention of an emotion or an act. The thing that drives the emotion is more important than the emotion itself. Don’t get carried away. For similarity in this quote in tech, find the why, its important, go a level deeper. The question ‘what’ is already answered; you need the core intention to build something, to solve something. Without the intention, all software is slop. ## Read [From words to worlds with spatial intelligence](https://drfeifei.substack.com/p/from-words-to-worlds-spatial-intelligence)- This a refreshing read, a positive take on the LLM thing. Its true, the words might have limitations, we don’t even know what true VLLMs are capable of, they can’t actually recognize the physical space, and what interpretations can they perform on it. - The one thing that stuck with me was the use of LLM in education or learning. It is ranked or stated at the very end, indicating that LLMs are going to be least used to reform how we learn. If that is true, then I can take a relief. If they can’t change the way we learn things, I am a happy and satisfied person. It helps me use my brain rather than I handing over my neurons to it. [The AI Wildfire is coming and its going to be very painful and incredibly healthy](https://ceodinner.substack.com/p/the-ai-wildfire-is-coming-its-going)- A beautifully written comparison of natural phenomena to the current situation. - The reset of nature as wildfire is really needed in Tech for AI to seed its place. It would burn the hype and leave behind it some mess and blessings that people who struggled here will clean up and reap the benefits. - I don’t have much to say, a lot to say rather. I recommend just reading it, it gave me calmness, it’ll be just fine. AI is not coming to burn you, don’t worry, but there will be a new resource at our disposal. You know it, it has its pros and cons, the AI chat. [Parsing integers in C](https://daniel.haxx.se/blog/2025/11/13/parsing-integers-in-c/)- Its again one relatable post. The author is pointing out that he saw a problem. Parsing and robust handling of integers in C. I love python for it. I wonder how is it developed on top of C then. If C is worse than python for handing integers, how is Python working so well. - cURL, that library man! The author and the creator of libcURL or cURL the tool is a legend, he is a gift to the developers and the world. The library is much more than a http client. It has laid so many ground works for making the ecosystem of working with the web and APIs coherently and without causing any confusions. - This post highlights the presence of parser for string to integer conversion in cURL as well as cURLX libraries. It handles them in a more robust way than the typical standard C libraries. [Recommendations for getting most from technical books](https://sebastianraschka.com/blog/2025/reading-books.html)- This is some gold advice on consuming technical content in general I think. Not just books. It could be a big blog post, video or any course. First consume it without distraction, try to understand next, try it on your own, try it again, think about what you want to do, do it and repeat from 2 again if you fail. [How to stop having FOMO as a software engineer](https://www.yacinemahdid.com/p/how-to-stop-having-fomo-as-a-curious)- This one is a hard hitting banger. yacine writes some banger posts. shed for a minute all the other’s expectations and look deep, look at all the stuff that truly brings you a sizeable amount of joy in life. - Yes, just bury all the expectation for once and do what you truly want to. I think this weekend, I would build a TUI for GCP Cloud Logger. Because I want to. No expectations. I don’t want to livestream, things get in the way, I don’t have the right set of mental clarity to write golang posts, I won’t for now. I’ll do what I love write now, things are overwhelming me. SQL, LLMs, VLLMs, Transformers, Vibe coding, PDFs, Rust, everything seems to sucking out the joy out of me, they are all great things, but I can’t handle them all at once. One thing at a time, and for this weekend, its TUI for GCP Cloud Logger, for sure. [Why I stopped trying to be a great engineer](https://supremecodr.medium.com/-56094c225549)- This is a wholesome post. Just read it, its so short. - It conveys that taking a break from the normal chore, brings a fresh perspective and unclogs the brain. It helps your brain get out of the rut, it shows you a new hope, it fires a kindle of hope and curiosity. I guess, stopping to code, or writing some code if you haven’t been due to AI, just do what you wanted to do, but something or the other kept bugging you. Because when the heart and mind are in sync, impossible becomes possible. [The quite power of SQL](https://blog.sturdystatistics.com/posts/sql/)- This is true. After all, sometimes, not all new tools are worth using all the time. Sometimes, maybe often times, we need to keep it simple and use the good’ol trusted tools, the simplest ones just like SQL, the dc calculator, how LLMs are using the existing tools to create wide possibilities. [How to demo](https://thorstenball.com/how-to-demo/)- Some great advice on how to demo. It has some good catches. Keeping it shorter, slowly moving to the solution, not banging it in the start. Let the viewer understand the problem in its entirety then develop the intuition to your solution. - Adding pictures add value, but I think it should be a concise representation of what is changing or what the crux of the presentation is, not just adding visuals just for the sake of it. [On doing things](https://jalexine.github.io/on-doing-things.html)- I don’t know nature is hinting this week to me to just do things. Right, I’ll do it. Sometimes ideas come when you stop trying to force them. - True. You have to just let it sleep with you. We say right, let me sleep on it, let it absorb, because the unconscious mind always keeps ticking off, it keep finding paths to connect, firing right set of neurons happen at a shower, at a walk, maybe just before the sleep. Just breathe, and let the world be it. Your in-action won’t cause it to fail, but don’t make it a excuse to never do it. ## Watched [How Martain was written](https://youtu.be/EXD3b6OLtsg)- I haven’t read Martain, want to. But nonetheless, this is an inspiring talk, as a writer and a programmer I am compelled to write a story now. This might be a turning point for me. If he can, I can too. - It appeared to me at the right time though, November in middle of NanoWriMo. Perfectly adding up to my writing streak and building confidence. [Building a terminal wizard in BubbleTea](https://youtu.be/Gl31diSVP8M)- This is a great video to understand Bubble Tea framework in almost 20 minutes. I was pumped after this to finally decide to build the TUI for GCP logger. Finally it will be done this weekend. [How OpenAI Atlas is built](https://youtu.be/jxsA_185lMI?si)- Atlas is a chromium wrapper right? right? Nope, its a little custom separate entity that is powered and laid by chromium but it doesn’t look and feel like Chromium. - I don’t know anything about what Theo said, but if he is saying its gross. It might as well be, I can’t comment if I don’t know it. ## Learnt Visual LLMs: Show then, don’t tell. - If the VLLM has visual recognition, prompt tuning will only work till a point, you then have to show it, not tell it. - Text LLMs are bottlenecks due to that exact reason, they can’t see. But if you add reasoning, it opens up a possibility. But if you power a VLLM with reasoning (Gemini 2.5 Flash) it is a superpower that you have honed it. I think this is barely explored territory and needs more eyes and experimentation. Golang BubbleTea TUI Framework - It has Model, View and Update as a core principle, which is the ELM Architecture style. - It fits this style of developing the TUIs so well. You have some state to display, you present it and you update it. Simple. You can build almost anything with this principle. Would be trying out to build GCP Cloud Logger TUI. ## Tech News [OpenAI releases GPT 5.1](https://openai.com/index/gpt-5-1/)- It looks like a general-purpose and more tunable model. Looking forward for the snitch bench test. [Google release Code Wiki](https://codewiki.google/)- Wow! This is a Holy Grail of code wiki. It can generate tons of things from a given github link. Very good point to start understanding a codebase if you want to contribute to. If you its hallucinated, you can prove it by running the code, simple It was a whimsical week, the writing just gave me everything I needed in a week. Peace and Clarity of actions. I am in the right mood, at the right time. Just by spending one hour each day, I become a better version of myself every day. Good to write this way. Hopes to continue in the next week and year too. For more news, follow the [Hackernewsletter](https://buttondown.com/hacker-newsletter/archive/771/) (#771st edition), and for software development/coding articles, join [daily.dev](http://daily.dev/).Leave a commentShare

Techstructive Weekly #62

Sat, 04 Oct 2025 00:00:00 UTC

## Week #62 It was a great week. I didn’t knew I would say that at the start of the week. But hell yes, a promotion, finally. A hope at the end of the tunnel. Hard work pays off in the end. I kept believing and finally able to bring smile on my loved ones. I quoted last week about home, a family, is not something that is by blood, rather it is the group of people who strive for each others success, for uplifting each other, when one falls down the other roots for him or her, when its harsh time, everyone steps back and stays together. And I was able to experience just that. The true value and meaning of a family and life. Not just my family. I was able to believe that the people I work with are also a family, after a long time, I was able to feel that I belong somewhere else too, I was appreciated. My work was worthwhile, the fruit of efforts are sweet, especially when you don’t expect it. I am quite happy and in a good mood. I am not able to quit this endless loop of promising a youtube livestream but keep on failing week-after-week. This week, let’s get it right. ### Quote of the week > **“**You weren’t rejected. You were merely redirected**”** > > — **Elaine Welteroth** Yes, I’ll go with this for the week. I did everything for 2 years. Same problem, same person. But things change, conditions change, you fight, you are put down, you try to get around it, but rejected, the journey seems over until suddenly you are automatically redirected to where you should have been. You get your validations and your result, you feel good, you move on. Simple, don’t remain stuck, don’t ask why? answer the questions what and how. You are in the situation, you have a problem to face, you can’t question nature, why it was me who was given this problem, that is not in your hand, simply answer what and how do I overcome. That puts you in way better position than grieving over the problem. ## Created Have stopped the SQLog for a while this week, was busy with other work and plus it was a festive season, so some festivities and longing for a deep mind reset. - [SQLog: Create Table with SELECT](https://www.meetgor.com/sqlog/sqlite-create-table-from-select/) - I found it interesting to populate dummy data from existing values. - The select statement is so plugable, it just gets into everything. ## Read 1. [The beauty of Programming](https://www.brynmawr.edu/inside/academic-information/departments-programs/computer-science/beauty-programming) - I love this post, it propagates the love of programming really well. - It shows why someone is so driven and addicted in a good way to programming, it feels like he can create anything. Someone might say, i just change characters in a file and make things happen with the machines. > you can be God. On a small scale - I love the problem from the Gaussian sum. > It’s still hard to explain what can be so fascinating about beating your head against the wall for three days, not knowing how to solve something the better way, the beautiful way. But once you find that way, it’s the greatest feeling in the world. - Yes this is the quote I live for and die for, it is the thing that gets me up from the bed each day. 2. [We traded blogs for black boxes and now we are paying for it](https://tekhne.dev/internet-resist) - We are paying for it, true, cannot stress it enough, we are just letting the humanness get away from humans. AI is doing the heavy lifting so that we can do the laundry. - Social media is just slippery slop of AI generated mess - Being authentic and miserable is more important then ever, feeling human like emotions is the need of the hour. 3. [An opinionated critique of Duolingo](https://isomorphism.xyz/blog/2025/duolingo/) - Duolingo is a great app. But it is just gamification of user attention. I tried learning Japanese, and I can’t understand the language as a human, as a repeating process it is nice, but no, it doesn’t click - I am using it for learning chess, so far it looks good, its like a textbook way of learning things. Nothing personalised or very sophisticated. But the UX is cool, the interaction and user feedback is smooth and top notch. - It is a engineering marvel, but not on the human-like-learning path. If it can do that, it is a great app. 4. [SQL Heuristics: ORs are expensive](https://ethanseal.com/articles/ors-are-expensive) - Wow! ORs are expensive. - I had never thought about it that way, would like to dig a little deeper here and learn more about the maths here. It hasn’t clicked all the bells for me yet. 5. [Write the damn code](https://antonz.org/write-code/) - I am just 101% on this. I want myself to start the code, the dopamine hit, the shit show. But AI is there to help, don’t use it as a one-shot thing to create a software wand, it is a great peer, a great co-pilot, you still need to know what you are doing. - Vibe coding is basically delaying the shooting yourself on the foot problem. - Get involved and get into the zone. 6. [Failing to understand exponential again](https://www.julian.ac/blog/2025/09/27/failing-to-understand-the-exponential-again/) - This person is reputed and he says LLMs are growing exponentially. Is that true? Maybe. There is value in LLMs true, it is hidden behind the hype true, the hype is infuated, true, but calling it exponential value, hmmm, maybe not. 7. [Advice I wish I knew as a junior developer](https://raheeljunaid.com/blog/advice-for-developers/) - This is valuable. I would add, a few more points. - Don’t hold yourself back in diving deeper, if you find something interesting, go in all the way. You’ll find yourself in a much better place then others. - Just do things, there is no reason to avoid things. Just patch one library, because you thought it was bugging you, just implement a markdown parser from scratch just because you feel the need to learn the process, don’t hold yourself back just because you’d think people would judge you, they will judge you anyways. 8. [Why hiring only senior engineers is killing companies](https://workweave.dev/blog/hiring-only-senior-engineers-is-killing-companies) - Juniors with AI tools might be worth more then seniors, because of their adoption and open-mindedness. 9. [The simple habit that saves my day](https://alikhil.dev/posts/the-simple-habit-that-saves-my-evenings/) - I need to think about it, I somehow waste my evening lying down and consuming news and tech articles. That is not wrong but I need to get the true dopamine by creating something, I end up in the same cycle of exhaustive slump. ## Watched - [Cursor AI Foundations](https://youtube.com/playlist?list=PLuI2ZfvGpzwCEXrl_K6bW5OqNpZq3HkMa&si=NmxQoNkfB8-0zCZP) - It was refreshing to watch the tool that I use day to day. I don’t like much of its usage. It causes me anxiety. But knowing its capabilities and limitations is not harmful. - I like the way it explained, the problems with AI and its sloppiness. - [Cursor designer builds an OS with cursor](https://youtu.be/TQhv6Wol6Ns) - This is wild, I cannot believe this, it was created by AI, half baked but still. I think it takes time to clean up the vibe-coded mess, often then not, I have to re-read and re-read until I scrape it off completely. - But this just doesn’t feel natural. It might be but I am not convinced. It just feels a high effort thing, not as the author described it. “Just two prompts”, “completed in an afternoon”. I am not sure that is true, could be skill issue on my end but cannot comprehend at the moment. - [Harshad Numbers](https://youtu.be/dgwevhEykWQ) - This is cool. Harshad, meaning happy numbers, they result in full integers when the divided by the number itself form the sum of its digits. - Mathematics is so infinite, this inspires me to write code and think about problems from mathematical perspective. Yes, sometimes, that gets in the way of the actual problem, but on the leisure side, its fun to play with math. - [Computerphile: Machine Code Explained](https://youtu.be/8VsiYWW9r48) - I haven’t seen such a simple explanation of Assembly Code and Memory in my life. The comparison of register as memory blocks, pigeon holes as memory units holding volatile memory. - I want to read [this book](https://yurichev.com/mirrors/machine-code-for-beginners.pdf). This is such a great visual of how memory and computers work. I think in the age of AI, shallow knowledge is no good, we need depth and low-level details, when vibe code slop fails, real grug brain developer has to roll off his sleeves and dug deeper. - [Cluade 4.5 is the best coding model in the world? Till when](https://youtu.be/uZBjVeyiYkk) - Claude 4.5 dropped, maybe its a 0.5x improvement, is it? Maybe but definitely it has reduced its behaviour. That is a good and a bad change. - Good because obviously, it should work in friendly way, AI and security right now are two separate circles, bubbles if you will. - Bad because now we it doesn’t quite do what it is asked to do, it might avoid the things that are not ethical, but it could have side-effects where it could reject a query which is genuine and not align with its interpretation of ethical. Double click to interact with video - [Claude made its own language](https://youtu.be/BiFt4vw1-uM) - Why? Really? It goes to show that people are ready to waste money for no reasons. And AI is far far from replacing developers. - Cursed programming. Cursed vibes. ## Learnt - SQL Using SELECT to populate and create a table. - We can use SELECT statements to create and populate tables immediately, It is usually a way to test or duplicate tables in a existing database. - However the SELECT statement won’t copy the constraint, indexes and all the schema of the table, it just copies the column data, the structure and not the logic behind the data. ## Tech News - ZAI releases GLM 4.6 - OpenaAI launches Sora 2 Models are getting launched but not having a significant bumps in their capabilities, which is a good sign for humans. But these labs are just going crazy. They are putting AI in any possible thing they can, social media, if that was not a worse place to be in, developing anxiety, loneliness and pressure, adding AI to it is going to doom the peace from humans.

Techstructive Weekly #61

Sat, 27 Sep 2025 00:00:00 UTC

## Week #61 A bit of slow and disappointing week on a personal note. I tried my best to be a backend developer but was not worthy of being one, I was stranded as a product developer. I am not saying I hate being a product developer, it just gets too menial and boring once you know the limitations and the quirks of a product. Apart from the grill, I learnt a lot about tokenization. I also continued to write occasionally about SQLite in the SQLog, three more entries. Wasn’t able to livestream due to guests and was unwell for the weekend with cold. Read a lot as usual, still keeping out from the doom-scrolling spiral. It is helping me on the mental level, but need to find something in life that will push me without looking for other things constantly. Maybe my work is not helping me here, its barely keeping me in peace, need to find a switch as fast as possible, the job market is not what you want to be in. Let’s see being conscious and specific about life is getting important and need to take some actions to lift my spirits up. ### Quote of the week > *“A home isn’t always the house we live in. It’s also the people we choose to surround ourselves with.”* > > — From the book “The House in the Cerulean Sea” by T.J. Klune I completed reading the book and felt deeply satisfied and cozy. What a book, hits hard, hits home. The way it presents dull and boring life into a beautiful narrative for a transformation. The lessons of empathy and kindness, being a human is so important in today’s age. Sorry, this is not a book review, but getting into the specific of this quote is grounded in the story. The main character finds his new home which once he thought would just be a part of his job to do the needful for a month. A home is not something you just live in, its a relation with the people you make, the care each other makes, the choices and memories we make is what make a home, family. ## Created Continued to write more about SQLite, not quite continuously. I had a break on Sunday, then got back after one day gaps. So only three articles for the week. But keeping it strong and sturdy. - [SQLite SQL: Many to Many Table without RowID](https://www.meetgor.com/sqlog/sqlite-many-to-many-table-without-rowid) - [SQLite SQL: Collate Column Modifier](https://www.meetgor.com/sqlog/sqlite-collate-column-modifier) - [SQLite SQL: Create Temporary Table](https://www.meetgor.com/sqlog/sqlite-create-temp-table) ## Read 1. [I don’t want to code with LLMs](https://blaines-blog.com/I-dont-want-to-code-with-LLMs#footnote-ref-1) - This is the best of the lot. Nailed every point. - Coding was never a bottleneck, communication was (is). - Yes, it can do trivial task, but that is not 20% of developers work. - Its still bad at complex stuff. Vibe coders hit a ceiling after a while. - Reviewing is worse than writing it yourself. - AI is a tool just like IDEs, you are not losing or missing out on it. - When did learning deeper and low-level knowledge get uncool or not important? It hasn’t, its just hype, hiding form us the actual pillar behind the rise in information accessibility. 2. [A New kind of Code](https://registerspill.thorstenball.com/p/a-new-kind-of-code) - I like the analogy of glue code, but I don’t like reviewing code. Its skill issue on my side, yes, but I can’t stand reading code, I want to see it, I want to feel it. Generating with a button doesn’t give me any excitement to push forward to get things done. - Maybe I am naive, but I don’t like that button. Sometimes yes I do use it, sometimes forced to, but I like to keep nuts and bolt and prepare it myself. I am learning still maybe that is the reason. Can’t disagree to Thorsten, who can write 2 books about crafting programming languages and interpreters and compilers. 3. [How I a non-developer describe a developer](https://anniemueller.com/posts/how-i-a-non-developer-read-the-tutorial-you-a-developer-wrote-for-me-a-beginner) - This is so funny. I can feel this. Documentation, who writes that. - Tests, what are those? We push -f to prod - By the way, here’s the rollback script if the production is doomed. 4. [Processes and Threads: Planetscale Blog](https://planetscale.com/blog/processes-and-threads) - This was cool actually. The interactive elements really make things so clear. They remove the textual-ness in the blog and add a depth to it. - There is a clear explanation of program, process, ram, fork and low level details that are quintessential to a developer, any f-ing developer. 5. [Be careful with Go struct embedding](https://mattjhall.co.uk/posts/be-careful-with-go-struct-embedding.html) - Yikes, this hurts sometimes. Its just like SQLite, but without guardrails. Golang tries to be too lenient with the inferring the values for a struct and create a bag of unexpected behaviours. 6. [Artists are losing work, wages, and hope as bosses and clients embrace AI](https://www.bloodinthemachine.com/p/artists-are-losing-work-wages-and) - This is a serious topic, I have stopped using AI to generate thumbnails for my posts and whatever artistic form I used. Firstly it looks and feels dull and un-energetic. - I do respect artists, this might be a disrespect for them to use their knowledge and wisdom without giving them a credit. - I feel bad about the writers whose style and wisdom is now in-grained in models like GPT, Llama and Claude and others too. 7. [Brace yourself](https://thedailywtf.com/articles/brace-yourself) - It is always that damn comment. Developers write code, run it, if it works, fine, but if it doesn’t work, then debug it (printf debugging) and fix it and leave, we don’t test. 8. [One last id](https://thedailywtf.com/articles/one-last-id) - Man! SQL is tricky sometimes. The number of abstractions people create in the dialects is so jarring. It really breaks stuff. ## Watched - [Tokenization from scratch from Andrej Karapathy](https://youtu.be/zduSFxRajkE) - What a beautiful piece of content. Archive and store it in a museum. The depth with which he explained it, the low-level details, the pythonic bits, is so fun and contagious to watch, and feel. - I learnt a few tricks about interaction with LLMs and understood certain quirks. This could give a intuition for why certain LLMs won’t be able to give good completions for certain tasks. - I also don’t quite liked the Sentence piece tokenization logic. But I can see where it could be probably come handy, in PDFs for example, the scope of sentence is well defined. In arbitrary piece of text on the internet, it might not be. Double click to interact with video - [Tokenization in C from Tsoding](https://youtu.be/6dCqR9p0yWY) - This was another great livestream like tutorial. The depth with which he communicated and came up with the solution is what helped me get better understanding of tokenization. - This is an example, why learning from first fundamentals is still cool. The ability to learn and explain with confidence and comfort is remarkable quality for a developer. This is also I am still thinking about doing livestream and being able to develop something from scratch. Double click to interact with video - [What is a Tensor? A beautiful intuition and question and answer based explanation](https://youtu.be/k2FP-T6S1x0) - This is a great piece of explanation. The question from Richard Feymann are so deep, provoking and sensible. The way the author explained and questioned his own thinking is really great. I liked that way of teaching. - The direction bit and the animation also helped a lot. Nice editing skills. Double click to interact with video - [I hate myself more for seeing this, than I hate javascript](https://youtu.be/7bvBVBy_CrM) - This is gross. I really hate seeing this now. I can’t bare this. - I mean, javascript is a good language, but why people just used it and didn’t improve it. The author developed it in a week, weren’t the industry leaders a bit mature to make it better? The Java developers have fixed things from it, but the thing that was copied to Javascript is still ain’t? This goes back to the meme of pillars holding the bigger stones. Double click to interact with video ## Learnt - How BPE or Byte Pair Encoding Transformer works - It basically finds the most frequent occurring two characters and groups them as a new token (apart from the individual token / characters) - These pair of word is added to the vocabulary or a lookup reference for mapping the character set to the id of the token. - Then it does the above iteratively until the most frequent pair has count of one, we can limit that. Then at the end the vocabulary developed will be used to encode and decode the content (text) - There is a sweet spot of minimum token size instead of converging to the entire dataset for maximum tokens, it can compress a lot of information about a single word that it might not be able to understand it. - This is a classical example and limitation of why LLMs can’t do math or count how many r’s in strawberry as strawberry might be a single token and it can’t get enough data about the characters in just one token. It has a lot of knowledge to get that number but just sheer looking at the tokens, it can’t. - The other example of adding less tokens is that we might have a lot of tokens for a given content and we might run of out context window to process the request. This is true for languages other than English, LLMs aren’t good at that the earlier ones especially as the tokenisation for those was not optimized or had fewer tokens due to selection of dataset. - SQLite - Create temporary tables, the temporary table is actually stored in the temp location file and in a separate database. - If you had two tables with the same name one as the primary permanent table and other as the temporary table, if you just refer the table name, the temporary table will be preferred over the permanent. You will have to use `main. `to actually refer to the permanent table. ## Tech News - [Postgres 18 released](https://www.postgresql.org/about/news/postgresql-18-released-3142/): - This has some cool features like generated virtual column as default generated columns. - [Planetscale makes Postgres Generally Available](https://planetscale.com/blog/planetscale-for-postgres-is-generally-available) - This is a big one, they are going all in postgres. - [Github launches Copilot CLI: Another entry for the Agentic CLIs, oh god!](https://github.blog/changelog/2025-09-25-github-copilot-cli-is-now-in-public-preview/) - We have Claude Code, Gemini CLI, Codex, Warp, Amp, Cursor agent CLI, and oh my gosh, this list is getting too big. Developers are using it, this is going to create a IDE-like war in the near future, if developers exist. - [Improved Gemini 2.5 Flash and Flash lite models](https://developers.googleblog.com/en/continuing-to-bring-you-our-latest-models-with-an-improved-gemini-2-5-flash-and-flash-lite-release/) - This looks like a good improvement, reducing the number of output token is a good sign from Google. They are not looking to increase their revenue, they are taking a hit for it. - [ChatGPT Pulse](https://openai.com/index/introducing-chatgpt-pulse/) - Not sure this is a right step for OpenAI, they are going to far with this one. - [Meta releases open weight coding model](https://ai.meta.com/research/publications/cwm-an-open-weights-llm-for-research-on-code-generation-with-world-models/) - This is a bit cool, it has almost the intelligence of Gemini 2.5 Thinking with a lot less parameters (32 B) - It also has a different and quite novel approach for a coding model That’s it from this week. It has been a harsher week. We are finally into the last quarter of 2025. Ah! That ended pretty quickly. LLMs have just consumed me.

Techstructive Weekly #60

Sat, 20 Sep 2025 00:00:00 UTC

## Weekly #60 Another great week, consistently learning SQL, developing another streak for solving problems on FreeCodeCamp with Javascript and Python, reading instead of doomscrolling. I have been learning about SQL since maybe 10 weeks now and finding myself in a good position, for the past 4 weeks I have written posts every day and it has helped me get back to the technical writing flow. On the work side, it was a fun week, tinkering with a system that is already working decently and wanted it to get the most of it, turning the bits and knobs of a system is so cool. This weekend, I think I would be simply starting a golang project to get tokenizer or something from scratch. I just want to relax and code something from scratch just for fun and learning purposes. Let’s see if I can livestream it. ### Quote of the week > "The first principle is that you must not fool yourself, and you are the easiest person to fool." > > — Richard Feynman I have been fooling myself that I know stuff. Until I started to learn SQL, and boy o boy, I feel like I don’t even know a drop in an ocean. Database is a vast field. Such is life, there is a lot to learn, once to learn something, you get a little egoistic. That is where you have to wake up and be grounded and humbled, realise that what you know might be temporary or is already known or created by some other human. Respect it and move forward to gain more and provide your experience and guidance to others, the same way the people have done to make you learn it. ## Created Wrote more blog posts on SQLite and Relations: 1. [SQLite: PRIMARY KEY column constraint](https://www.meetgor.com/sqlog/sqlite-primary-key-column-constraint) 2. [SQLite: PRIMARY KEY table constraint](https://www.meetgor.com/sqlog/sqlite-primary-key-table-constraint) 3. [SQLite: Foreign KEY table constraint](https://www.meetgor.com/sqlog/sqlite-foreign-key-table-constraint) 4. [SQLite: One to Many Relation with Foreign Key](https://www.meetgor.com/sqlog/sqlite-one-to-many-foreign-key) 5. [SQLite: Self Join Relations](https://www.meetgor.com/sqlog/sqlite-self-join-relations) 6. [SQLite: Many to Many Relations](https://www.meetgor.com/sqlog/sqlite-many-to-many-relations) 7. [SQLite: One to One Relations](https://www.meetgor.com/sqlog/sqlite-one-to-one-relations) This makes it the 32 posts in total, have been writing daily for a month, and could see a lot of confidence and new energy surging within me. Writing really liberates the soul, it gives purpose. ## Read 1. [My Favorite Postgres 18 feature: Virtual generated columns](https://tselai.com/virtual-gencolumns): - I agree to this, there are pros and cons of both. Stored makes write heavier but are read efficient. Virtual makes it write easier and read heavier. You have trade-offs, you need to decide based on the computation that impacts how you want the column to be generated. - I don’t like the notion of JSON flattening in Postgres. Postgres is not a database that would be ideal for that kind of data. I know there are tons and tons of support for JSON, but tables and JSON, I can’t bare it at once. Those two are just separate entities for me. Maybe they are useful in one-off values, not not much. Switch to NoSQL if you have that lengthy data. 2. [Boring is good](https://jenson.org/boring/) - Yes, this post summarises the current trend in LLMs well. The hype is about people adopting to the value of LLMs, but soon they’ll will realise that they were on the wrong path (one example is they thinking, LLMs could replace developers) and then we will settle on the thing they are good at. I know, throwaway code, temporary code, and simple stuff that you know you need to do, you know how to do, but not worth the time to manually type it in and craft it. - I am really excited and positive about the SLMs, the small language models, I want to use it to just be a google search but simple and not ripping out my entire project into a react and python boilerplate mess. 3. [Work Hard, have fun, go home](https://www.bonnycode.com/posts/work-hard-have-fun-go-home/) - If you have fun you can work sustainably, else no matter how trendy or shinny the technology you would use, you won’t endure it long enough on the verge of burnout and the feeling of unsatisfied will crumble your efforts. - Maintaining balance is key, but flowing with the excitement and energy is also helpful not always but can be done to improve the fun and memorable parts. 4. [Building a lexical analyzer from scratch in C](https://devlogs.xyz/blog/building-a-lexical-analyzer-from-scratch) - This actually cleared up how to write lexers from scratch. I was wondering if that was bunch of if-else to parse each tokens, but we do have to group the kind of tokens and then write specific conditions on how to parse them. Now that makes a lot of sense. - Worth exploring more by writing my own markdown parser, even adding more features and syntax. If your soul screams to write your own flavour of markdown, let the muse take over you. This month or one day, not this weekend though. 5. [Creating a static site for all my bookmarks](https://alexwlchan.net/2025/bookmarks-static-site/) - I am very much this kind of person, I want to access my bookmarks and the linkblog too. - I am developing it, but can’t make it polished enough to be usable. I had created linkblog.netlify.app. This is work in progress, anyone can add but I would like to add the authentication, but that again creates a friction element, this all constraint bother me then. 6. [Python can open web browser for you](https://koaning.io/posts/python-can-open-a-webbrowser-for-you/) - Python -m webbrowser - That is so cool, can’t think of other automation that can be done here. - Neat and handy 7. [Representing Graphs](https://thepalindrome.org/p/representing-graphs) - Edge List: List (tuple) of nodes and optionally weights. - Adjacency Matrix: Matrix of each node with the all other nodes, really great way to describe the graph, has everything that is needed to understand about a graph. - Adjacency List: Map of each node with a list of nodes that are connected with it, if weighted, then can add a list of tuples representing the node and the weight. A little tricky but the author says this format is the most used and is optimal for general use cases, so seems good enough. [![](https://substackcdn.com/image/fetch/$s_!5Jm3!,w_56,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff8b68cf8-d3f4-42f6-b8dd-cccde036005f_720x720.png)The PalindromeRepresenting GraphsHello there…Read more6 days ago · 51 likes · Alberto Gonzalez](https://thepalindrome.org/p/representing-graphs?utm_source=substack&utm_campaign=post_embed&utm_medium=web) 8. [Go Struct Alignment: A Practical Guide](https://medium.com/@Realblank/go-struct-alignment-a-practical-guide-e725c8e1d14e) - I have read this and it makes sense, a bit wired but nice. Writing structs should be carefully planned, so just add the largest ones at the top and cram all the smaller ones thereafter. The rule of thumb to follow if you have any memory-heavy or scarce use case. - Handy functions like Sizeof, Alignof, Offsetof are used to get the total byte size, memory alignment requirement, the field start position of the struct or any fields. Use it to craft the proper and perfect structure by tinkering and aligning. 9. [Myopic Focus](https://thedailywtf.com/articles/myopic-focus) - Wow! Fantastical and Tech stories, are my love. - Gist of the story is people are very myopic(viewing in very hindsight) about system and think of the existing workflow as ritualistic that makes it harder for others to make it better or even improve the quality or realibility. - The story goes like > A new developer chops replaced fragile ID pool logic with robust UUIDs, eliminating inevitable crashes. But their manager fixated solely on declining unit test numbers, demanded rollback. It made the project a bit risky to sustain, but the fault is not of developers. Its the myopic view that causes it. 10. [What AI chatbots are doing under-the-hood, LLMs from scratch part 1](https://www.gilesthomas.com/2025/08/what-ai-chatbots-are-doing-under-the-hood) - This is a great series, I am reading more about it this weekend. LLMs are things that produce logits, which is like a list of probabilities from a vocabulary and for each token the highest probability token should have been chosen, however it is random the temperature determines how random the selection is, 0 means choose the first(highest probable, no creativeness), 1 means choose any random one. Between these 0 and 1 you can experiment to find the sweet spot for your needs. 11. [Magical Systems Thinking](https://worksinprogress.co/issue/magical-systems-thinking/) - I love this post, it hits home for me - Systems should grow from the simplest possible solutions and then branch off from the possibilities and situations. - Creating patches to the existing systems will only survive for limited time or none at all. > Sundar Pichai estimated in late 2024 that over 25 percent of Google’s code was AI generated; as of mid-2025, the figure for Anthropic is 80–90 percent. - The comparison to AI slop is great and fits well here. Vibe coded mess is no exception to systems thinking. We have seen LLMs want to patch and keep on patching existing code mess, we want to start from scratch, the urge is right, but we worry about whether we will lose the progress? The progress is fake progress. - The code LLM produces will always be throwaway code, as it is no subconsciously written, call me philosophical, but code is art and it needs attention form a human soul in order for it to work, not technically but overall in order to complete its purpose. > NEW SYSTEMS CREATE NEW PROBLEMS’ and ‘THE SYSTEM ALWAYS KICKS BACK’. As systems become more complex, they become more chaotic, not less. The best solution remains humility, and a simple system that works. 12. [The Sad, Sad world of Tech Blogging during the Era or Technological Stagnation](https://freddiedeboer.substack.com/p/the-sad-sad-world-of-tech-blogging) - There is nothing really ground breaking in mobile phones that is true, in LLMs too this seems to be getting closer to. The 90% closeness is achieved, the 10% will be like a forever process, people making predictions of AGI by 2027, and all that hype-crap is non-sensesical hypothesis. - Does this mean, tech world is plateaued? Maybe not but unless we have something wild, nothing is going to change marginally just like gpt 4 and 5 bump, they know it but want us to feel like its a jump [![](https://substackcdn.com/image/fetch/$s_!no2m!,w_56,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fbucketeer-e05bbc84-baa3-437e-9518-adb32be77984.s3.amazonaws.com%2Fpublic%2Fimages%2F0bc5fd66-6f8a-4d34-add5-3eff35a4e30e_512x512.png)Freddie deBoerThe Sad, Sad World of Tech Blogging During an Era of Technological StagnationI have a piece for Vital City arguing that, to succeed in the negotiation and compromise that will be required to build a ton of new housing, YIMBYs have to stop engaging in convenient and false caricatures of who opposes new construction. Check it out…Read morea day ago · 120 likes · 54 comments · Freddie deBoer](https://freddiedeboer.substack.com/p/the-sad-sad-world-of-tech-blogging?utm_source=substack&utm_campaign=post_embed&utm_medium=web) ## Watched - [David Heinemeier Hansson (DHH) on Rails World 2025: Opening Note](https://youtu.be/gcwzWzC7gUA) - I can’t fathom here, how did we end up backwards? It takes some thinking to deploy to prod, yaml manifests, and what not to deploy a simple API, how the heck people are accepting those? - It is like a egoistic culture to have complexity and assume that it will work, but they are only adding complexity upon layers of complexity that developers will have to clean up themselves. - Rails is doing a phenomenal job in the developer ecosystem. Double click to interact with video - [OpenAI dropped another Codex](https://youtu.be/j9wvCrON3XA) - I agree to this, it is just a slight improvement to the GPT-5 experience. Not much here. - I kind of hate to admit that free users can’t use the web codex, the cloud instance of the agent. So, basically free users are devoid of the code experience of the GPT models. - Google might win that race. Gemini CLI is so cool, they also have Jules. Double click to interact with video - [Finding and Exploiting Buffer overflow password cracking in C](https://youtu.be/W1okWEzGAGo) - This guy is producing great stream, bangers after bangers. This one not quite remarkable but enduring nonetheless. I will vouch for persistence over hype and fuss. Double click to interact with video ## Learnt - Javascript Tidbits (TILS) - Rounding to a float using `toFixed(n) `this will round off to n digits of a decimal/floating number. Python has round(), Javascript has toFixed, kind of confusing name but fine. - Math.floor() to convert a decimal to an integer, this I used it to perform integer division. I come from python and `5//3` would give `1` but Javascript `//` is a comment literal so need to use Math.floor instead. - SQL Logs - Learned to use Foreign keys and create - One to One relations - One to Many relations - Many to Many relations - Self referencing relations - In SQLite, foreign key constraint is not enabled by default, you need to explicitly set it as `PRAGMA foreign_key=on`. SQLite is very flexible, though it sometimes feels like a double-edged sword, so flexible by default, strict only when said, kind of good, but one can easily shot on the foot. - I tweeted about it [here](https://x.com/MeetGor21/status/1967818642768470447). Generated columns are quite new. This was added in: - SQLite [3.31](https://www.sqlite.org/releaselog/3_31_0.html) (in Jan 2020) - PostgreSQL [v18](https://www.postgresql.org/docs/18/release-18.html#RELEASE-18-UTILITY) (in Sep 2025) - MySQL [v5.7](https://dev.mysql.com/doc/refman/5.7/en/mysql-nutshell.html) (in Oct 2015) ## ## Tech News - Java 25 Released - How can one miss this update? - [Compact source files & instance main methods (JEP 512)](https://openjdk.org/jeps/512) - Finally the boilerplate is reduced in Java, one less thing to hate Java for. I never hated it for that, but never mind. - OpenAI releases Codex, model, CLI, and cloud agent, or everything - Hhh! How many products Open AI has with Codex? I know the name is nice, but move on with it and be creative, the o- series was less confusing then this mess. - [Meta AI live demo of glasses that went bad](https://techcrunch.com/2025/09/17/meta-unveils-new-smart-glasses-with-a-display-and-wristband-controller/) - Atlease they tried, that’s what I would say, they aren’t faking that it works like magic unlike others. - [Notion Launches Agents in 3.0](https://techcrunch.com/2025/09/18/notion-launches-agents-for-data-analysis-and-task-automation/) - Every product is becoming something with AI. - [Duocon 2025:](https://duocon.duolingo.com/) - This is an App I am admiring for its UI and UX, so clean and intuitive - Addictive design is a different thing, but its utility and worth using - I was awaiting the [Chess feature since May](https://blog.duolingo.com/chess-course/) and it finally launched this week. Can’t wait to up my chess skills That was a heck of a long week. Lot of tech things happening. It will get even more in the next couple of months, as it is October and there are already a lot of developer related things, conferences, hactoberfests, and among other things.

Techstructive Weekly #59

Sat, 13 Sep 2025 00:00:00 UTC

Week #59

Another productive week, a lot shipped, almost all critical bugs fixed, the launch looks great. Wrote daily for another week about SQLite/SQL. Adding up 24 posts. Read it here. Generating a lot of ideas, getting back to journaling, reading instead of consuming videos and doomscrolling. It is getting better day by day. I have completed my yearly goal of reading 12 books. Still 3 more months to go, would be almost at 15-18 books.

I am healing. Another week, of not writing much production code, but helping a lot of code go to production by pointing out bugs, testing, debugging and importantly defining the problem and its possible solution. Coding is not the only thing about being a software developer, there are all of those elements to that make one better at the craft.

Looking forward for a month full of SQLog, might live stream this weekend after again 2 weeks of procrastinating about it. Have a bit of ideas, and the inspiration that the coding from scratch is not dead yet, learning is still alive, people might feel value in what I do.

Quote of the week

"The quieter you become, the more you are able to hear."
-- Rumi

Ah! This is what I have been doing, just hearing what nature has to say. I am reading and journaling instead of consuming endlessly. This puts me in a thinking and attentive zone, rather than the consumption cycle that drains energy and causes anxiety. Being quite here doesn’t mean not doing, it just means to step down your brain power and think about the things happening, without getting distracted, lets embrace boredom.

Created

SQLOG Entries for 7th September to 12th September
- Continuing with some week of Learning SQL probably 8, but 24 days of consistently writing a blog post. Feels refreshing and accomplished. It really feels good to look at the number cracking up and the blog filling up, the sudden increase in likes, followers and sprinkles of curiosity to learn more.
- Writing about the things we learn is underrated.

Read

What would you do with 52 hours of discretionary time
- This is a fresh perspective, I haven't taken that analytical approach in measuring the number of hours spent in weekends. That's quite a lot, 52.
- Would be spending some time to plan and think about it, it really had me thinking now.
I wrote a journal entry for everyday in 2024
- I had journaled for 1 year in paper, 2 years on my phone, and the last couple of years, have lost that habit. Life gets in the way, your habits change, but this I think I have to keep, it used to give me clarity and feeling of accomplishment.
- It has been exactly a month since I started journalling again, pen and paper, a tiny little book, around 50 words, I would cramp 100 in that little page. I love it, its like a glimpse of a quarter in a tiny book. I am also thinking of getting back to obsidian for digital journal, just because its accessible and searchable. This article I will give credit for bringing up searching and doing some analysis on the things I have already written, that would be exciting to find how I change or my writing changes me.
The last programmers
- There is so much wisdom in this post, I feel like quoting a lot of things, but the ending nails it "The parts that have always mattered, really. Understanding people". There I put it, no fuss, no bait, just facts.
- Maybe we are the last generation of coders that type code by hand, and push to production. We are seeing the hands taken over by AI slop. I call the code slop, it is. If not reviewed by a human, it is garbage, well not entirely but nothing short of saphegetti, legacy, I-don't-want-to-touch-and-read like code.
- The author is hinting at the transition from developer happiness to user happiness. I don't like that but it's the harsh truth, the hard pill to swallow. No one will be a true nerdy developer anymore, they all would be average vibe coders. Only the best among them will be truly nerds. Will that change from now? I don't think so, people can use computers now, but in earlier days they don't use to. People (in India) at least don't know the proper usage of AI, its actual working, that's why AI-bros exist. Sadly they would sell these AI as their product but that is I think would be a skill, to steer AI in doing what you want, and for that you need to understand what they want. Pretty long rabbit hole but worth thinking about as a developer.
Package Managers are evil
- This is a fair take. Absolitely, the left pad example from Theo is one such thing. People just keep on adding packages/libraries without thinking much, in pre-AI days that was the problem.
- But now, with AI, it can spit out code like anything. No need to worry about managing packages, but eventually it will be producing more code which is a liability. AI produced code might be fragile, very like todo: authentication coming soon, like code. If not tested or reviewed, can't trust it.
- He is right on all the points, Golang is batteries included, and clearly defines what a package actually is, its just a folder. You can import anything from the folder. Except only if the functions or structs are capital case (annoying at times, but fine). Having some rule is better than having none. Javascript failed to define a rule, and NPM is a mess. Golang doesn't have a package manager, it just manages itself.
- I also find python dependency management like javascript to be honest, but a little better with terms of completeness. Since people can mess up on the web pretty easily, the things to mess up with Python have a less surface area. If you are aware of what happened to PyPI several times, you know what I am talking about, Its common to manipulate a source of truth and people might find themselves in all sorts of trouble, they would have never imagined. With uv I think it is moving to a better place, but still the core of the problem is from the too much of flexibility, which is fine, and needed even. Python doesn't needs to be like Go.
When the Job search becomes impossible
- This is gold. It might not look as relatable to everyone or even every time for those who relate to it now. But for me, who had spent 4 months finding a internships, 87 applications to land an interview, I can surely say there is a hope at the end of tunnel.
- I am saying this to myself again, maybe in the future, if this job search turns out to be the three stage one, impossible search, next to impossible search, wired search. Maybe sometimes, life wants you to show something different before you see what you wanted to see, and that might even amp up the feelings after going through that path. Maybe it was not the path for you, you were all on the wrong path, maybe it was a long path, maybe it was multiple paths, and you'll have to keep switching paths, who knows?
- I started to slow down and learn things, If people say you are not worth hiring, don't give them a chance to say that. I started writing about a skill, every single day, even if it might be very small thing I learn, I write it down in the blog. I don't care if no one sees it, I don't care if in the era of AI, blogs are not searched, I don't give a shi* about SEO, its all doom and gloom. But I do keep my knowledge to myself, it just takes a moment for someone to see my work in these days and recognize the skills and if not I still have the skill, If can't prove the skill then I need to improve.
- "Rest is all about space. It engages purposefully with serious boredom".
- The above quote is true, I am not sure, if people are getting aware about this doom-scrolling thing, but its real! I got sucked into it in the month of June-July somewhere and it felt like my brain was rotting. I immediately took a step back, paused all notifications and started reading instead of scrolling. I read books, I don't have money to buy but you know, I love fiction. I completed around 6 books in the past 2 months, and wow what a feeling to bear. Too much dumped here, let's write a post about it. Thriving in the boredom or Reading is better then doom-scrolling.
Blogs used to be different
- This is so well written, the conversations have been lost, it feels all clickbait and flow where the trend takes me, instead of the flow or my curiosity.
- Blogging in earlier days, I am speaking from 2018-2020, I knew it used to be very much tied to google, I never read a lot of blog post at that time, so I can’t comment, but can feel the lack of social media would make it really amazing back then.
13 Reason why SQL has to GO
- https://www.infoworld.com/article/2335455/13-reasons-sql-has-got-to-go.html
- I don't like this take, nor I think is grounded. Its looking like a AI slop or maybe someone just living in the land of NoSQL.
- "Tables don't scale", what is the meaning of it? If that's the case, then collections also don't? Right? I bet PostgreSQL and MySQL aren't using tables, then it means that MySQL doesn't scale, come on!
- I don't want to react to this clickbait, I already know the pros and cons of SQL vs NoSQL, SQL isn't going anywhere, and so is the good old MySQL. Relational Databases are the nuts and bolts of web, and without them, industry can fall apart pretty quickly. Unless MongoDB keeps them happy :)
IT manager needed tech support because they had never heard of a command line
- Can we all laugh here? I mean, yes its not needed, but at least heard of it! SKILL ISSUES.
- Or Maybe he was rich, he had macbook since his career began. I didn't had a laptop, I found the command line on my android phone! Also discovered the world of Vim, because Nano and Vim was the only option, and I choose the Command line and Vim came along.
Become the person who does the thing
- https://www.fredrivett.com/2025/09/10/becoming-the-person-who-does-the-thing/
- Yes, the world right now feels like everything is pushing me in a direction that I wanted, in my favor. For real! Journalling, job searching, embracing boredom, learning and locking in, and the most recent thing since almost a week, building a exercise routine.
- I also believed physical fitness is not important as mental well being. But I am coming to realize, there is no mental fitness if the body is not fit. At least it should be moving. By working remotely I was sucked into sitting and consuming junk/content. Feeling lethargic and exhausted all the time, last weekend, my brother set me up for an exercise, I felt energetic after that. I have been doing minimum 20 minutes each day since and keeping away from the vicious consumption cycle.
- Side by side, I am also on a writing streak of 24 on SQL and SQLite on my blog here with sqlog. This all is pushing me to get better and be the person I want to be. This article helps me validate everything I am doing, I don't need validation for the actions I do, but a human is a social animal, he needs resonance and some level of accountability or validation for what I am heading for is the correct place or not.
Forward vs Reverse Proxy
- This is cool, the examples, make it really interesting and relatable.
- Forward proxy acts on the behalf of the client
- Reverse proxy acts on the behalf of the server
- It actually makes sense and helps understand better.
Just use SQL they say
- Use tooling around it, there are SQLC tools for go and I think even for Python. It will help you write type safe boilerplate code in the native language as you write the SQL query.
- I think the author is right, but unaware of the tooling that exists around it.
The AI Trap: Why do they quit coding
- Programming isn’t dead. Programming is not just about coding, its about problem solving.
- The joy of problem solving would still be there. It makes a little harsh on people to quit as some of the casual programmers who tried to earn money would definitely quit.

Watched

Python: The documentary
- Wow! Just wow! From abstracting the language for making it friendly to running the world. It went from 0 to 3 pretty quick. Its the language spoken by LLMs
- Dropbox was surprising, the whole stack is Python! how would the frontend by python, maybe server side rendered, but that is really sick. The scale of that working makes me think again, is python slow? or people make it? by adding unnecessary complexity, django scales!
- I read a few post from Amir about Claude code and AI stuff and was about to disagree, but he is a creator of Flask, and oh my god how many frameworks? I took a step back and read some of his blogs again and it changed my mindset. Maybe LLMs are cool, maybe we need to be better programmers, we need to define our problems better, that’s what I was missing. This guy is a legend.
- The pyladies moment made me cry. Not gonna lie, this was really relatable. I also find loneliness and under-confident when given an opportunity, maybe its just me.
- Guido is such a humble and mature person. He took the leadership as if he was like the care-taker of the language, I mean he is the creator of course, but most of the people just take the money and go away. He lead the development forward and made the language what it is today.
- I cannot relate to 2 to 3 switch, but can understand why it might have happened in the first place. Writing an API for the first time is cool, no constraints, just buggy code to make it work, but then you realise oh, that is a bad code, here and there, and it becomes necessary to refactor in a way that it has to break the backwards compatibility.

The fastest sorting algorithm: Radix Sort
- Simply, we take the bases or digits in the number and sort them in buckets, the memory might be high, so that’s why its done per digit to avoid billions of buckets being created for each number.
- I like the approach of floating point numbers, the guy really thought it through, was really clever of using a format that would sort the number as a digit. Absolutely fantastic mathematics in this.
- Will try to implement in golang to get a feel for it.

DNS Lookup from scratch
- This was a refreshing post, it made me think,
- Programming is not dead yet.
- I have lost the motivation in streaming because of AI, no one would watch someone code something for hours, but this is bringing that back.
- The joy, the delight, the satisfaction of coding your own DNS from scratch in C, elite programmer. I would recommend subscribing this guy, full of energy.

Print function in Python
- Wow! I thought print was for stdout, but it was meant to be written to the file. So cool. I though write was the only option to write the file, but this is so crucial to know.
- Print is a very versatile function. Sometimes, it makes me wonder, is python really a magical language.

Learnt

Javascript
- 2d Arrays
  - Creating and Initialising an 2d array
  - arr = Array.from({length:n}, () => Array(n).fill(val));
  - Here n is the dimension and val is the value to prefill with
  - For instance if n = 2, val = 0 then arr = [[0,0],[0,0]]
- Set difference
  - To find the difference in two sets in JS
    - > s1 = new Set([1,2,3])
      Set(3) { 1, 2, 3 }
      > s2 = new Set([2,3,4])
      Set(3) { 2, 3, 4 }
      > s1.difference(s2)
      Set(1) { 1 }
  - Or from an Array
    - Create a set, for each array
    - Then from each array find the element that doesn’t exist in the other set
    - This will give the non-present element in both the arrays
    - Combine the set to get a array of elements that are different in both the original arrays
    - > a1 = [1,2,3,4,4]
      [ 1, 2, 3, 4, 4 ]
      > a2 = [2,3,4,5,5]
      [ 2, 3, 4, 5, 5 ]
      > s1 = new Set(a1);
      Set(4) { 1, 2, 3, 4 }
      > s2 = new Set(a2);
      Set(4) { 2, 3, 4, 5 }
      > d1 = new Set(a1.filter(n=>!s2.has(n)));
      Set(1) { 1 }
      > d2 = new Set(a2.filter(n=>!s1.has(n)));
      Set(1) { 5 }
      > [...d1, ...d2]
      [ 1, 5 ]
- String split
  - Javascript split is bit under-powered compared to the Python split
  - > s = "hello world, this is two spaced word"
    'hello world, this is two spaced word'
    > s.match(/\S+/g)
    [
    'hello', 'world,',
    'this', 'is',
    'two', 'spaced',
    'word'
    ]
    > s.split(" ")
    [
    'hello', 'world,',
    '', 'this',
    'is', '',
    'two', 'spaced',
    'word'
    ]
    >
SQLite
- There are 4 column constraint namely
  1. NOT NULL
  2. GENERATED ALWAYS
  3. DEFAULT (Not really a constraint)
  4. UNIQUE
- TABLE Level constraint
  1. PRIMARY KEY
  2. FOREIGN KEY
  3. CHECK
  4. UNIQUE
- Using combination values for unique constraint
- Understanding the difference between stored and virtual generated columns and comparing against the reading of those columns with time bench marking
- The check constraint is not a column level constraint, its a table level constraint as it can use combination of multiple columns for managing (updating or inserting) a record / row.

Tech News

Claude model degradation: I don’t use Claude much but it had a reputation of being one of the premium and high quality availability model which with this seems to be deteriorated.
Nepal reverses the social media ban as the protest turns deadly: Really a deadly situation in Nepal, I can relate as I have my work colleagues working there. Government is pretty unstable and it might create havoc for such a fragile nation which might add misery to the already poor state.

Happy Coding :)

For more news, follow the Hackernewsletter (#762nd edition) , and for software development/coding articles, join daily.dev.

Techstructive Weekly #58

Sat, 06 Sep 2025 00:00:00 UTC

Week #58

What a week, almost broke prod for half an hour! On Monday, can you believe it? I didn’t but that was a bug. Patched it and moved on. Had a consistent posting and learning sessions daily for an hour about SQL and SQLite. Feels good.

Had an situation where knowing javascript was a big deal and I don’t feel like writing javascript, it just doesn’t appeal to me, I guess I have to amp that up. Anyways, one more reason added for hating javascript, its unavoidable.

The only painful thing about Javascript, its unavoidable

Yes, back to SQL, I am absolutely feeling energetic to talk and learn about SQLite daily. I think we’ll be 100 posts in no time. I am thinking about creating and completing a project this weekend. Let’s see if we are able to do it.

Quote of the week

To live is to wrestle with despair, yet never allow despair to have the last word
— Cornel West

We all have problems, but we can’t let it define us, rather we will be defined on how we reacted to them. You can give up, but either you’ll be unsatisfied or below average. You have to go through the pain in order to bear the fruit, and even if the fruit doesn’t wag your way, so be it, if you enjoyed doing it (writing or reading in my case) then I believe that was worth it. Else what is the purpose of life? Who are we really and why do we even breathe. Just enjoy the thing you enjoy there is should be no measure for productivity when it comes to living.

Wrote + Created

7 More posts in the SQLog Series or SQL Blog
Updated my blog SQL codeblocks to support latest SQLite version 3.49 checkout here
A demonstration video explaining about the blog I created link

Learnt

SQL week probably some 7, I am loosing count, let’s forget the week count from now
- SQLite has a rowid as a default primary key for distinguishing each row for a table. If created a primary key, then rowid is an alias for that column
- Different modes in sqlite and how they get affected with the headers dot command.
- SQL is a relational algebra under the hood, select, relation, union, projection, intersections, difference, joins,product.

Read

The modern job hunt: Part 1
- This was a heart warming and hope-giving post. For a person struggling with jobs, finding a new hope is something that this post absolutely gave in abundance.
- The comparison of a walk to garbage collection is so funny. Job search as a searching algorithm.
- I also like the two different contrasting thoughts and way of thinking about the situation, from
  This job search is a hopeless, unending slog!
  TO
  Yes, this will end. Everything ends.
- This hits home. Truly a great post, Eagerly awaiting for a second part. Subscribed to the RSS feed only to know the feed isn’t updated.
An illustrated guide to OAUTH 2
- This made everything click, not quite everything. I want a better example. I will write a post where I would relate it to a much more simpler and broader example. The Chase bank and other finance provider is a bit not known to me, I want to make the example clear.
- But really the explanations in this blog are really sketched out and make a lot of sense.
Why Sync?
- I see a lot of sync things, like sync engines, local database sync, I even myself had this issue on blog sync with database and github repo.
- My manager hit me, a quick sync? That’s a different sync of course. Sync!
A programmers guide to stress
- Wow! What a banger of a post, what gold pieces of adivce
  we cannot get rid of stress. What we can do is change our philosophy and the way we see things.
- This blog advocates to meditate 10 minutes a day, it helps maybe. I need to try. But that comes after doing something, you act first, then if the result is not coming, you get anxious and frustrated if it failed, in that case, you need to calm down, you did whatever you could, but now you need to wait for it to resolve or learn from the failures.
- Getting anxiety from the current problems is true, I cannot keep my mind away. I need to breathe and let it go, will try over the weekend and continue if it makes me happier.
SQL Needed structure
- This is a bad take, What is more structured than SQL? If you are trying to get heirarchial data in SQL, you will be comprising redundancy. Like literally for the ease of the access, you will duplicate json blobs everywhere. I work in that environment and know the pains of those queries. I won’t recommend MongoDB ever to anyone. Just use structured data, it would be a one time investment on schema design, but it sustains and is much easier and predictable.
- It could have been other things like sharding, I would have agreed, but this nope.
Static Site enable a good time travel experience
- A really cool observation. Anyone can pick a commit hash and simply run a command to obtain the static files. Or better just checkout to the output branch and go back to the date. Its like a time machine.
- I love static site generators.
You have to feel it
- Feel it, Just completing tasks, merging PRs, lists of metrics improved won’t help, if you don’t feel it.
- Just creating for the completion is fine, sometimes needed, if you haven’t gone through the whole feeling, but not always, you need the drive to complete it in a sustainable way.
How to learn System Design
- System design, something I can see myself struggling a lot if in a interview. Need to start this approach of creating one my approach in designing existing system.
Authorisation explained
- A nice walkthrough of authorisation types, role based, attribute based and access lists.
Vibe Coding is Legacy code
- This is the realisation I had last week too. I used agent like the code based agents to write a script that I won’t care about, some code that I won’t care about after a while. That type of code is a vibe.
Reading with AI
- People are pushing AI in wired direction. By using that for activity like reading, you are defeating the whole point of reading. Non-ficition or ficiton, if you put down the effort of going through word by word, the context, the feel to the LLM, you are doing yourself some deservice.

Interesting Links

30 Minutes with Strangers

Watched

Database System: Relational Model and Algebra
- Prepared a detailed notes as I was watching
- data model - how to define the relation, rules
- schema - defining the specifics of the data model, details
  - relational (the primary kind, we only think of this as real dbs) key value - graph - document - column - array - and more
  - Initial idea: Writing cobol in 1970s to get data, literally telling the db to how to get the data. Not ideal as you don't know if that would be ideal way to get data in all cases, example artist and album, you write for each artist, find albums, but what if the number of albums by each artist keep growing and more people ar accessing different artists data together, so not sure how to get data for any query. In SQL however, we dont say how to get data, we say what and from where to get data
  - Relational Model
    - Structure
    - Integrity (constraints)
    - Manipulation
  - Its upon the database system how to query/mutate the data
  - Components
    - Database storage(bits)
    - Physical storage (pages, files, etc)
    - Logical Schema (schema, constraints)
    - External Schema (common table data)
    - Application
  - Everything below application is a db system
  - Relation is a set, relation of attributes that represent entities
  - Tuple is a set of attribute values in relation. Primary key, uniquely identify a tuple in a relation. Foreign key, related tuples(attributes) across relations. Constraints, conditions must hold for any tuple in a relation.
  - Relational Algebra
    - Select (where conditions)
    - projection (select with what to extract optional modify the selected values)
    - union (all, must have same number of attributes)
    - intersection (same but common)
    - difference (same but difference)
    - product (cartesian product, cross join)
    - join (natural, without params, common ones, same attribute name, can use params like using **on**)

SVGs are so cool
- I really didn’t knew SVGs can be that cool, you can scale, animate and even make interactive elements.
- I need to add some SVGs on my blog. This is absolutely inspired me.

Vibe coding are three types
- I know and care about the code
- I know the code but don’t care enough
- I don’t know what I am doing
- Developers, let’s focus on the first 2 and we should be fine, the 3rd one is a luxury as a hobby.

Tech News

Atlassian to buy the Browser/Arc Company
- I don’t care.
Warp release Code: the terminal based coding assistant
- This an another weapon added to the arsenal we are seeing in the arms race of the AI coding assistant. How many we have, I forgot the count.
Anthropic raises $13B Series F at $183B post-money valuation
- I don’t know what that means, but is a big thing according to the big tech bros.

Nothing happening for a few weeks now, finally there is a slow down in the models and all this hype. I want this hype to be minimum as possible for a stretched duration of time, as it helps learn and do things peacefully.

Happy Coding :)

For more news, follow the Hackernewsletter (#761st edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #57

Sat, 30 Aug 2025 00:00:00 UTC

Week #57

It was a fun week, really doubled down on learning SQLite and SQL. Shipped a bit of code and had fun creating improving metrics on the work side of things. Able to see 10% improvement in the things I have been working on the past few months. Really a good vibe week. I finally broke the barrier or imposter syndrome and whatever it might be called it while using the AI-assisted/vibe coding things.

Use AI for what’s it god at: Throwaway code

I actually started to understand the areas where I need and can 100% use AI tools like throwaway scripts, CSV/JSON dumping, data extraction scripts things like that. I used Cursor-Agent to do the most of things and I was happy at the end of the day, it nearly one shot it, I had to do a few tweaks here and there but was mostly correct, nothing like starting to write the script from scratch after it giving up. It really helped me save at least half to an hour worth of work in 5 minutes almost 3 days in a row. I felt good, I don’t felt like doing that chores so nothing valuable is lost in that.

I can focus on the other priority items. So, this was a valuable lesson for me to use AI effectively where it shines, in generating shit lot of code that nobody cares but I need to get the job done, generate a few data points from sapheggeti logs and piles and oodles of data. I don’t care if it uses a list comprehension or a C-style for loop, I didn’t even bother looking at it, here readability has 0 value. So use AI to generate plots, structured data from logs, benchmarks (review the samples first though), among the other things.

I can breathe in comfort, after knowing that I develop that intuition as a developer, those things are still valuable and evergreen even if AI can spit out code, there needs to be the one who can guide it in a way to get the job done **effectively**.

Enough of shilling and rambling, I am back on track with the excitement and fire back, will be recording a few videos and might do a live stream if time management is done right this weekend.

Quote of the week

"You may say that it was the book, but it's how you read a book that is most valuable, rather than any power it might have itself."
— Michiko Aoyama
(Author of What You Are Looking For Is in the Library)

I read this book over the weekend, and really found it stuck with the weekly things I did. For instance, the AI thing, it might be powerful but if we don’t know what it’s capable of really and where and when to use it, it feels worthless, it feels forced, it feels hyped (not gonna lie, its over-hyped though).

Created

Added Interactive SQL Code-blocks / Playground on my blog
Added RSS feeds specific to tags and post-types
LogSQL Entries for each day of the week
- I wrote a bunch of articles around SQLite specific things like IIF, random, unistr and unistr_quote scalar functions, the read dot command, and specific things about the version upgrades.

Read

Learning Peal in one day and the importance of building strong foundations
- Great article, it shows the author cares about his craft, I can relate to the feeling of learning a lot of things.
- Doing things faster can lead to cutting corners, that’s a good take. I think we can agree to that, we are loosing the ability to know something deeper. Learning is becoming easy, but not deeper, its shallow, brittle and hallucinated. Failure-based learning is more valuable then productive driven success.
Job Churn
- Hope, is what the article gives. If the history has thought us something, then it will repeat itself and we will have jobs disappearing and appearing again, in a different form and domains.
AI Slop in Open Source Contributions
- I agree to this post, its quite obvious that if the author of the MR/PR has just submitted a vibe contribution (without even looking or testing) he/she doesn’t even care about the project, its a red flag.
- I have contributed to a few open source projects and have been on the other side (contributor side) and I use AI to understand the problem, help me write a fix, but I do that in a worktree separate from the branch that I am working, I nit pick details that I think are important, it produces a lot of things, I examine each line and word carefully and have a lot of to and fro for understanding why it added it, usually the reasons are lame and on the lines of of course why not! I delete without thinking twice.
- I think people are against using AI as is and not as an assistance for helping you contribute and solve the actual problem, if AI had solved it, why even the maintainer would require you to create a PR and all the overhead in between understanding your thoughts on this.
OCR is Legacy tech
- Maybe it is, but not completely. There will be need of low cost easy document extractions. Yes, if the cost factor goes down, OCR would be obsolete.
- I work in these domain, and can say, a lot of things are changing quite rapidly, LLMs and VLLMs are taking the extraction industry by storm.
- Its quite easy to one shot an extraction with VLLMs given the intricated details might be missed by OCR and with VLLMs, its like a brain powered visual detector. It really understands the context and its knowledge really comes in the clutch to parse almost any type of document.
- Really excited to see where my journey leads in this development
Grok 2.5 is not really open source
- The License is not truly open source
- Its not open weights as there are restrictions on its usage
- So its basically weight available model, and you can’t really get any value from the weights.

Watched

Teej Devries on the Wookash Podcast
- I like Teej Approach, read the f-ing manual. Writing cover letter, I am not sure about the people might get negative impression of AI-written, but I can see the effort put in by the candidate or the desperation of getting the job.
- Going full time on streaming is really bold decision and he is really a great person to do just that, a real practical teacher. A true inspiration.

Please Stop using booleans in database designs
- This is great advice, I get the point, having updated_at is much better than having is_complete, etc.
- The enum example also makes sense. Its basically like having a breakdown of the field into required pieces.

Learnt

SQLite Week 6
- I learnt a lot while creating the bits and pieces in my daily logsql blog posts. Especially the details of the dot commands, scalar functions and new updates in the SQLite 3.47+
- The IIF is more rich now, as it can take in from 2 to N arguments, where N can be conditions and pair of conditions to get the value if the condition is true, this is new, the older one was 3 way condition, the first condition, the second as the true value and the 3rd being the false value (it can be nested). However this both are now in the same function, maintaining backwards compatibility. It also added an alias as IF making it easier to come from other databases.
- The unicode and unistr and quoting the emojis in SQLite. I am suspecting because of LLMs these two functions were added, as LLMs are generating a lot of emojis people might be tempted to store something in the database and use it later, so escaping the unicode and quoting it is required to make it work with the unicodes.
- The read and the output commands are really interesting, the former is even recursive, found a bit of easter egg with it today.

Tech News

Zed adds Gemini CLI as a support for A2A protocol
Claude for Chrome
X AI makes Grok 2.5 open source (not really, its just weight available model)

Happy Coding :)

For more news, follow the Hackernewsletter (#760th edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #56

Sat, 23 Aug 2025 00:00:00 UTC

Week #56

A simple yet rewarding week. Continuing the learning path of SQL, and taking it to a next level with consistent posting of log like posts, reading a ton of hackernews articles, researching about AI generated images metadata, and a lot of python code.

I will try to take a break this weekend, try to think what I want to see myself in the next 5 years, 3 years, and 1 year. I really want to think what I am and want to be as a person. Thanks to the post that I read this week, also will continue with SQL lessons or even take a course or certification.

Quote of the week

“Being born is probably the most difficult thing we ever have to do. If you can survive the ordeal of being born, you can get through anything.”
— Michiko Aoyama
“What You Are Looking For Is in the Library”

This is such a great quote. It shows the meaning in birth. If you read the book, you’ll understand it much better. But TLDR is that you are in someone’s womb, you don’t know what is it to breathe, to open eyes, to cry, to feel your body, but still you move out of the womb and you are born. You faced the most non-deterministic thing. Now everything after that will be simple compared to that.

Similar is the situation of suicide or death feelings, if you are ready to give up and die, you have gone through everything and ready to get over the most dreadful things of life, death. If you had that feeling, you are at a much high tolerant level, everything else feels small and you could handle it, just hang in there, something beautiful might be coming your way.

Life is about extremes!

Wrote

Anyways, lets see what I took under the creation section for the week. Back to writing after a while. I will write things out, no matter how silly or existing crisis it feels. It is my thoughts, no AI shit, just simple old school learning.

I am also sharing it on Substack Notes, Twitter and Bluesky

Read

How to stop feeling lost in tech
- This is a premium quality post. So much valuable advice for juniors and people who are all over the place and feeling overwhelmed or burned out.
- What do you want to be in like 5 years, 3 years, 1 year? That hit me like a truck. I never think about the future, but still there needs to be specific I enjoy and would want to keep enjoying and learning. I know that's a too generic AI like answer but that is where the real friction will arise and I think we will emerge out of it with a fresh perspective. After having asked a lot of questions and having a better idea about our likes and dislikes, thinking without any pressure just gives us the opportunity to truly see beneath us.
- I don't think the steps mentioned there are necessary, like I know this might be done for relaxing the user (taking leave in the middle of a week, going a entire day at the waffle-house or cafe). But the process is to the point.
- Listing down the goals and what I want to be like a tree like structure, no cycles. Breaking down one goal from 5 year to 3, then 1 and then month and then weeks. That makes it really clear and forces us to double down on our true "calling" or the muse if you think in that way.
What makes a good software engineer
- Curiosity to build the right thing, and being honest about it.
- Trust is super valuable, I have experienced it working remotely at my 1+ year stay at my current company.
- I often just do it for doing it, then I hit it with I need to know why this needs to be done and done correctly. Then things take shape, curiosity out performs every other emotion.
Go is still not good
- I can agree to only one points here especially the standard library swallowing exceptions part. That sometimes is little on the border of Go's philosophy vs the actual issue. Just read the f-ing manual would be the argument, but it kind of becomes too verbose and critical if mishandled.
Exploring EXIF
- A great post, diving into a specific tool for a lot of things.
- I think the media metadata is messed up, the software is not consistently adhering to the standards. Look at the blog post and the Apple photos have a lot more metadata then Google or any other photo taking software.
- I have to say the different of fields are in orders of magnitude more than the rest. How much additional info they cram into a single photo is bewildering.
- I really liked the post, it had dumping of data into sqlite database and writing queries to get certain things, how cute and interesting that idea is. My brain is running in all directions at this now.
- LLMs to write queries to get photos from albums with natural language and metadata without actually processing the image, is one I can hit straight off my head.
- I confirmed that it Apple who is pouring love into those photos. It probably has to do a lot of things in the background to get the metadata. Such a irony of tech today. Lagging behind the trend but way ahead of everyone in the core.
How AI writing supercharged an addictive pattern
- We'll have been there and done that. The initial WOW! to the glorious vibe crap that comes after 12 hours of prumpting.
- I know there is a lot of value in AI, I am not a AI doomer, but the things people are assuming like 10x dev, replacing a junior dev, an intern, I don't buy that both emotionally and mentally. It just hurts the morale of humans to say that. If that is the case, then just flip the lever and let the software world be in chaos.
- AI assisted coding can be a bit addictive I agree, that's why if there is some joy in my side project, I do start it myself and then ask AI bits and pieces to figure it out, I don't let it rip my project, I am still an advocate of chat based AI-assisted development. That friction of copy-pasta I can tolerate but not the shit-shoving of vibe coded mess.
5 Docker networking concepts that everyone should know
- This was a fantastic walk-through of one of the un-explained parts of docker. Surely there are exhaustive lists of commands to perform things in the docker networking, but very few dive into the why part. This article specifically did that and all the commands made sense.
- The bridge is the safe default, setting custom host is for advanced usage, avoid if possible, only dive if necessary.
- Custom networks makes communication clear between containers
  user names, not IPs.
Left to Right Programming
- If python looks odd, how about SQL? That is left to right but evaluated mostly from the right to left right? mostly I asking loosely here, though aggregates might be exceptions and other things I am not aware of.
  Programs should be valid as they are typed.
- Good points, yes python’s list comprehension is an outlier at it as it needs the last bit first to understand the whole context.
No AI is not making Engineers 10x productive
- See the copium? Too much bearish on AI now. Suddenly people realise they need to put breaks to the hype and focus on improving the craft before cracks start forming into dents and large holes.
  Making all your engineers feel constantly anxious about their performance is bad for your company. It will make your engineers not want to work for you. This is a recipe for short term thinking that will encourage engineers to max out bad metrics, like lines of code. Code review will get neglected, tech debt will compound, and in the long term the whole company will be footing the bill of those errors
- What a relief, to hear or sorry read it. Every word of this article is healing me. I will write a separate thought on this article.
LLMs makes us dumber in the longer run
The comfort we get when offloading our cognitive load to LLMs is bad for us. Cognitive load should exist, and if we reduce it too much – if we stop thinking – we can actually unlearn how to think.
- Well said. Consistent reliance on AI is and will distinguish some human from an honest caring human.
What's the point of vibe coding if I still have to pay a dev to fix it
- People are coping seriously on AI now. Not that AI is bad, it's not yet there to build software blindly and let laymen control the steering. Devin, you tried, unfortunately, we won't be moving forward with your application.
Why building my blog is more fun than filling it
- Its a rabbit hole. I have done it and seen people do it too. Waylon Walker's Markata (its a python plugin based ssg, so easy to extend and setup) and my static site generator (I am calling it burrow, inspired as a home for golang gophers). I used jekyll > markata and felt the calling for writing my own ssg and here I am.
- It feels rewarding to show the world what you have built for yourself and brag about it when someone sees it. It's like a garden, what is writing equivalent in gardening? Planting trees, that's how the blog gets populated, and adding different things to the garden is what is addictive and feels like a gardener.
Do blogs need to be so lonely
- Collaborative blogging, I think hashnode tried to do it with community blogs and co-authoring. I like it but I feel that it's quite rare to collaborate on shared interesting topics, a podcast or a video interview might be a better fit to show the communication. Blogs don't really fit that quite well in my opinion, but yes distinct parts of the blog could be collaborated.
Stupid things that work
- I mean why not? looping one billion times in javascript makes your computer warm, fine. I would rather install and run Android studio than touch javascript. These days, running a local model around 3-5 Billion parameters (on 8GB ram) could easily burn your computer not just warm it.

Interesting bits

Duck DB Interactive tutorial
Improve your CV
Deploy FastAPI with Docker and K3s
Ask LLMs what time it is
- Prompt
  What is the time?

Watched

Tokens are getting expensive
- Geez, that is a lot of money. Models are really a way to fool and cash out money from people, these AI labs have just turned LLMs to be money making machines.

Didn’t watch anything apart from this in tech, read a lot of things this week, so off out of social media for a while.

Learnt

SQL Week 5
- Learning the different output modes of output mode of a query like excel, opening the result in a editor.
- Using generate_series to generate interesting data points
- Understanding the whole function parameters and ways to use generate series which is a virtual table / function.

Tech News

Notion Releases Offline page support
xAI secretly drops Sonic Models directly into Agentic IDEs as their Coding Models
Google almost halts the maintenance of Pytype: Another project in the Google's Graveyard.

Happy Coding :)

For more news, follow the Hackernewsletter (#759th edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #55

Sat, 16 Aug 2025 00:00:00 UTC

Week #55

A clean week, it was a festive week. Saturday was Raksha Bandhan, so almost half of the day was spent in those rituals and getting in and out of the house and guests.

Sunday was mundane as I just sat back and took a break to detox the technical things I have been doing over that last week. Then hard hitting monday came where I was frustrated with AI vibes, that left me a headache and sore throat for the next day. Took a leave for the day and was feeling better on Wednesday, cradled a few more vibe coding sessions, this time more focused on my code expertise and drove the LLM to do certain things in certain limited ways, and it worked better.

The tuesday bit is what gave me a little relief, I just layed back and slept for the most of the day, I had minor headache and throat was sore in the morning part. I just didn’t pick up my phone or the laptop till late evening and felt calm and human again. I think I can say humans need to have a digital detox day once a week or months to develop their brains again in this AI-dominated world.

Regarding learning more SQL, I took this week off. I think I need some time to accept all the things I have learnt and start a fresh next week. This is something I read somewhere that your brain needs time to retain information, too much information dump makes it harder to remember and connect things. Slow down the input and let the output process itself.

I almost completed the before the coffee gets cold series, the 5 book series, and I haven’t complained an inch so far. The meaning conveyed is heart touching, and it has given me a reason to exist and life among this frustrations and melancholics of life.

Quote of the week

“If we let everyone who dies to be a cause of unhappiness, that would mean people are being born to become unhappy, but people are always born for the sake of happiness“
— Toshikazu Kawaguchi from “Before your Memory Fades”
LINK

I read this book, and man! this quote hits home in the end. It just changed the way of my thinking about life and my loved ones. It gave me a reason to exist (among AI doom and gloom), it gave me a reason to be myself and choose being happy over sad.

If life gave you an option to choose anger over empathy, always choose empathy, this is also on the similar lines. If you have choice to be happy or sad, be happy, because that is what we are here for. Why be sad and live life like a zombie, like a AI chat bot without a true purpose and inner fire? Live life like a smiling and let that happiness spread and remove negativity.

Read

Grug brained AI developer: LLM Appendix
LLM worst when grug not know answer and hope LLM figure out.
this important distinction! many grug not understand!
- That made me laugh hard, not gonna lie. Truly relatable
AI Efficiency? Give me a break
1. People are seeing it, people are coping now.
2. AI is good, but not great. It has its value but the hype people created is 200x the value. Please, use AI I don’t refrain it, but everything has its moderation and area of use. AI for ART is pointless and gives existential crisis, refrain, I am not sure where it could be used without a thought blindly. In programming? yes to some extent in generating throwaway scripts or prototypes.
1910: The year where the modern world lost its mind
1. I could not relate but can feel the dredge in me when Machines might be taking over humans. Similarly AI is like taking away some jobs, but probably creating more to manage and architect around them, just like engineering was booming after the industrial revolution, AI management might be a field that would be full of potential.
I tried coding with AI, I became lazy and stupid
1. I have tried it too, I use it, but yes I have found myself in the author’s place.
2. It helps but only in that fine moment, after the work is done, I don’t feel rewarded, I don’t feel fulfilled, I didn’t learn anything, barely any dopamine.
3. I am better polishing off my skills
Why I choose OCAML as my primary programming language
1. This is huge post, the person lives and breathes OCAML. There are so many golden pieces of resources and thoughts. I would bookmark it for later detailed review when I think would be a right time to learn OCAML.
2. One interesting thing is that OCAML is a blend of theoretically established and practically used language. Which hardly any language does it that well, the ecosystem and the community too looks rich and supportive. These things must be true as these observations can be taken from years of involvement in the project/community.
Cross Site Request Forgery
1. Such a nuanced and detailed post about how CSRF is a tricky problem to counter. Why they exist and why one should be careful in building web apps.
2. AI generated slop code could probably cause CSRF issues but its very hard to say from a humans perspective, since the generated code is often hard to debug and pinpoint the issues as it would take considerable time to read through it.
Curate your own newspaper with RSS
1. In this constant hole of doomscrolling which I think I was falling for the last couple of weeks and finally got out by reading books. fiction books. It is important to make these kinds of apps and rss readers, as to be mindful about what we consume and at what rate.
2. In those days (2014-2017), I used to read newspaper after coming home from school. Those were just a 10 minute glance at the first 2 pages, but majority of time was spent in the sport and science zone. I loved reading it, that now makes me wonder, that even little and moderated content is enough for the brain to sustain. This never ending social media has made us brain want more and more, so we need to bring that fun and content (satisfaction) in content back.
How React works behind the scenes (under the hood)
1. Wow! What a post, I can happily say, I know something about React now. There are a lot of moving parts. I thought it was just one step compilation, but man there are layers of compilation happening in various formats/data layers I think.
2. The first one is JSON, the components and the apps are converted into JSON notation and then used to construct the DOM, that is fascinating, based on the reference to different components, it can decide which elements to render or re-render. There is a graph created so that it becomes easier to distinguish and make a hierarchy of the app. The virtual DOM as called is like the graph that helps in re-rendering, but the heavy lifting is done by the Fiber tree (which is a lower level abstraction) that does the actual replacement or rendering technique algorithms.
3. The process is so quick that the magic is not even noticable, but yes that is one rabbit hole to dig into.
Hugo + Obsidian + Git Pipeline
1. This blog gave me a idea to simplify my static site generator, instead of syncing with the database from github, I can just simply use a sqlite file and sync to the database an vice versa. Right now the problem is on the inconsistency in github content vs the database, so after having an in-memory or local file that can reside at any time on the repo, it would be easier to pull, push changes to the remote repository.
How I code with AI on a low budget/free
1. I need to try Cline, claude code router and qwen models. I have been sleeping on local hosted models, nit entirely though. I have a low spec laptop, 8GB RAM, so can’t use a model beyond 2 or 4 Billion parameter.
2. Kimi also has an API and a free credits, might as well give it a shot to use claude code for quick improvements to some projects.

Watched

Rewriting SQLite from scratch: Database School, Glauberg Costa, CEO of Turso
- I have been keeping an eye on turso, I really like the product, but now after watching it and hearing the mindset and approach of the company, I feel I need to get involved and contribute to this.
- This is such a great learning experience, I had cloned and ran the project locally last week and would find quirks and new features to implement next week.
- In a nutshell, turso started as a SQLite fork, but they realised the embedded replicas, the server and all the other things around sqlite were just hacks and they needed something better to support it natively. And hence Turso (limbo) was born which is a sqlite written ground up from RUST. I was amazed to see almost all the features are supported out of the box in this repo. github.com/tursodatabase/turso

Wishful programming
- This is so intuitive. I think Mitchel Hashimto too recommends this. He just prints out the things that he wants to get finally out and works to reverse engineer how it can be programmed. Mind blowing idea but quite simple to follow.

DuckDB in 100 seconds: Fireship is back?
- In 2025, we haven’t seen any 100 seconds video from Fireship, is this some sort of a cope to AI slop or really we think he was in crisis.
- But anyways, duckdb is cool, I haven’t tried it, it does use column based instead of tables or rows, which is kind of cool.

Learnt

Working with AI uses 2x my brain power, I am better off writing myself for complex or very detailed things I need to work on
- I was coding in Cursor for experimenting a few things on the pdf extraction and plotting certain tables and grids. LLMs are very nasty when assuming a lot of stuff and doing somethings on their own. And it didn’t miss a chance to do it there, to extrapolate the coordinates of the grid on the page for columns and rows.
- Also the frustration crept hard when the cursor was extremely slow for GPT-5 and even any model for that time frame on Monday and Wednesdays.
- I did so many iterations and finally gave up since it was quite frustrating to review the AI generated slop. I thought to myself, I could have done that myself in 2 hours rather than vibing it for 4 hours. Lessons learnt, never vibe code ever for tiny experiments or fixes.

Tech News

GitHub CEO resigns and GitHub is no longer independent after that
- People have raised concerns that it might be because of AI, but that really makes GitHub in a position of bother. It might even get backlash and people might start finding alternatives.
Claude 4 Sonnet now supports 1 Million tokens of context
- This is huge jump, it was 200K token previously, making it a 5x more context.
Google releases Gemma 3 with 270 Million parameter
- Such a small model and perfect for low end devices, but not sure about the performance to intelligence ration. Obviously it will be small, but if its faster and as neck to neck as llama 1 Billion then that makes it really the best one out there in this range of parameters.

Was a bit of digital detox week, 2 days, I just did nothing but stare at the ceiling and out of my window. This gave me a mental reset and some good vibes to just live. Hoping to crush next few weeks.

See you in the next one!

Happy Coding :)

For more news, follow the Hackernewsletter (#758th edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #54

Sat, 09 Aug 2025 00:00:00 UTC

Week #54

What a week! LLM models all over the place, gpt this, gpt that.Anyways, I find myself in a better place by avoiding AI in some places and using at its full potential at certain place, it took some time to realise it, but its kind of working now.

I don’t use AI to build the side projects like one shot, but use AI to help me understand certain topic, do something myself and converse with it about what it means, and what gaps I have in it. Using AI (cursor) at my work to speed things up and staying calm and letting things happen.

This week I continued learning more SQL, for the past 3 weeks, I have decided to learn SQL, just because I am curious and lo behold found a bug that can be fixed in SQLC, patched it, created a pull request and done. Started a series for solving SQL problems on exercism SQLite course. Finally back on livestreaming. Finally a good week, after a while. Feeling like a human, after dreaded existential crisis due to AI maybe.

Quote of the week

"The self is not found in comfort, but in the friction of becoming."
— Rollo May, The Courage to Create

I think I had been stuck in a cycle of consumption, I think I am coming out of it, by creating, I have not written much. This week I am thinking of writing blindly. I have recorded a couple of more videos for SQLite problem solving, and planning to add more over the weekend. There are a few hackathons that I think I would be participating, and would have a few projects out of it to showcase it. That was friction and that gives me the dopamine to wake up in the morning.

Created

Exercism SQLite Track: I started a series of solving SQLite problems on exercism and have plans to solve leetcode 50 sql, CS50 sql problemsets too. It will take time, but I would have learnt SQL then no other resource but practise and also upped my communication and teaching skills.
- Hello world and Local environment setup
- Two Fer

OpenAI’s GPT-OSS Models review for 20 billion parameter model
- Video and article
Open source contribution to SQLC
- Bugfix on DROP/ALTER/INSERT table values for case-sensitive table name
  - For issue > #3936: Bug on SQLC generate for queries relating to dropping, and update queries having case sensitive table name whereas, the select, insert and delete queries are case-insensitive to table names.
I finally livestreamed after a month or two about Golang project.
- Pocket clone, simplifying the initial prototype. No AI or vibing, just me and the editor. I use AI autocompletion, the next word autocomplete (calling it copilot level like thing pre cursor era)

Read

The fastest way to detect vowel in a string (Python)
1. Wow, this dude just found 11 legit (almost 13) ways to detect vowels in a string in python.Such a great depth, the benchmarks feels so intuitive as why each way performs the way it does.
2. Here are all the ways it did it
  1. For loop: Simple, readable. Fastest for small strings
  2. C-Styled for loop: Uses or comparisons, but surprisingly much slower
  3. Nested for loop: Totally exhaustive, but slow
  4. Set intersection: Clever and clean. Great when strings are long or vowels are sparse
  5. Generator expression: Pythonic one-liner. Reasonably fast, readable
  6. Recursion: Functional but inefficient. Crashes on long strings
  7. Regex search: Shockingly fast. Calls C-level code internally
  8. Regex replace: Works but inefficient. Doesn’t short-circuit
  9. Filter: Readable but wasteful because it processes the whole string
  10. Map: Similar to filter but slightly better
  11. Prime Numbers: Extremely creative. Maps characters to primes, uses GCD. Way too slow to be practical
3. Would like to do something in Golang, it sounds so fun that I can’t stop thinking about so many ways to do so trivial things.
Don’t tell engineers what to do?
1. This post, hits me and in a good way. It basically helped me get out of the self constrained mindset that I was doing wrong, instead I think I am doing my job. That is to listen and implement, the moment I raised objections or pointed out a few things, I was turned down as a kid on a shop. I am not embarrassed about it, just that the mindset in which I am working might be a little old or startupy. I am not saying I am perfect, just that there is a way to address engineers and their concerns.
2. Everything in the post that has been said, is totally relatable and worth reading for me. I read and felt a deep satisfaction that I might be just right and move on with the things.
Python’s pass by value and pass by reference
1. This is one hell of a reason, Python gets a little more confusing and less friendly.
2. TLDR of the post is that if you pass a immutable variable/object to a function call in python, you need to return it back from the function (if the function modifies those immutable objects). Because the object is immutable it won’t get updated inside the function, it will be created a new, so we need to assign it to the modified version when the function returns.
3. But for mutable objects, the function can modify it and we are passing it to the function, so the object will be updated.
The art of saying yes: How do I do so many things
1. Totally relatable post, I do a lot of things, because I am curious and excited to learn about it.
2. Starting small, building momentum, building a habit, is really what got me into this exact newsletter, one brings other, I kept writing one week after other and here I am writing the 54th edition. Not all weeks are fair, the same intensity or energy, some are dull, frustrating, just barely pushing past, but there are mostly the weeks where I know what I am doing, why I am here writing the post, full of curiosity, eager to learn more, and break prod.
Flashy, Fancy shortcuts aren’t always suitable
1. Using or to handle mutable default values in Python functions can create unexpected new objects when passed an empty list, breaking the link to the original. The standard if None check is more reliable and readable, ensuring the original list is modified correctly.
100 Rust exercises to learn Rust
1. So good. I will be using this to learn rust within a week it seems. I skimmed through a few posts, and found it really nice and interesting to go through them, clear and nice examples.
Stop building AI tools backwards
1. AI should augment, not replace, human learning through effortful retrieval and collaborative iteration. Buttons that auto-do tasks deskill users and erode trust, prioritize interactions that spark learning.
How to become a confident software engineer
1. Valuable advice and practical too
  1. Pick and learn a single programming language properly (like I am doing with golang)
  2. Write unit tests and CI (Ah, now it makes sense)
    1. I have not written serious tests in my 1 year long career, rest apart from the 4 year tech journey. Now it makes sense, there is a purpose to give me confidence to build reliable software.
  3. Make refactoring a habbit (encourage to do, but done or not is sometimes not in our hands especially for juniors)
  4. Pair with people (I do frequently whenever the need arise, and learn a few git commands and browser speed up shortcuts which help me learn more)
  5. Read books (I am currently reading Learning SQL to learn SQL in depth)
  6. Teach what you learn (doing it by youtube or articles I post)

Watched

You can just do things : Aaron Francis at Laracon US 2025
- You don’t need to be extraordinary just be out there to do things. Being kind, and expressing interest and curiosity is the way to get things done and build meaningful connections.

RSS is not dead
- RSS is not dead. Wow! I just had this vision of building a read later app a few weeks back, when the Pocket was deprecated. I want to build a combination of Read it later and RSS reader, a sync engine is something that might be used here if using cross-platform.
- Now I understand the conversation here, if someone is fetching the posts and storing metadata then it makes sense to have a paid solution.
  Bluesky post
- But the other point I am afraid of is the content scraping, is it ethical to do that? Like I know its not stealing, but crawling a legit site and using it on other platform is not good for link building for the author of the original article right? I am not sure, need a bit of thinking and ethical considerations.

The painful truth about startups
- OK, it was a harsh journey. Success is not overnight, its just a preparation for a overnight success. Uploadthing, round, picthing, and so many thing, and then finally T3 chat (chat thing, better chat), that made it.
- The salary looks like it will be life worth income. Man! india has a really low income-expense ratio.
- Just nerd out about the things I love, that’s what I am doing here.

Learnt

SQL Week #3
- String concatenation in SQLite
  - printf function that exists in SQLite. That is such a great thing to have, C like function, just drop the placeholder for the appropriate type and it will be a formatted string, neat and tidy.
  - || operator for simple concatenation. You can just use “Hello” || “ World!” to get a string “Hello World!”. It’s compact but for large number of strings, or readability, it gets a little tricky.
- I finally wrapped my head around autoincrement and sqlite_sequence table.
  - Autoincrement is a constraint that will force the newly inserted rows to use primary key id values greater than any existing or deleted primary key id values.
  - If the highest row created is deleted before inserting a new record, the sqlite_sequence comes in handy to fetch the max value of the primary key id across the table, since the value is stored in a separate table called the sqlite_sequence with table_name and the seq columns.
  - The sqlite engine decides to get the max of the seq value and the current max row id (it can effectively get it using B+ trees, as it will be the leftmost node or right-most node, however its stored)
  - I will be creating a separate blog on this, a deep dive on the various cases we can run into if the sqlite_sequence table gets altered.
- USE common table expressions in sqlite
  - This is a way to create a temporary table (like only valid till the query completes running) and use it in the main query(can have nested queries too)
  - The syntax looks like this :
```
WITH <temp-table-name> AS (SELECT something from somewhere)
SELECT something, <temp-table-name>.something from elsewhere 
```
  - This is something I studied while solving the difference of sum of squares and squares of sum problem in Exercism SQLite track.
- SQL’s IIF is equivalent for CASE WHEN THEN
  - The IIF is like an handy if else block to use when having nested conditions.
  - This works and looks neat for smaller expressions, like one or two condition max, after that its better to use CASE
  - The syntax looks something like:
```
SELECT IIF(something > 10, "YES", "NO") as answer FROM somewhere;
```

Interesting Links

https://github.com/zakirullin/gpt-go
https://blog.jetbrains.com/education/2025/07/28/rust-exercises-rustrover/

Tech News

Phew! can we start counting the number of models released this week, this was that was kind of week.

OpenAI releases GPT-OSS (open weight models under Apache 2.0 license)
OpenAI releases GPT 5 chat, mini, nano with and without thinking capabilities
Anthropic drops Claude 4.1
Google’s Genie 3
Horizon Alpha and Beta models on openrouter
OpenAI releases Study mode in ChatGPT and Google releases Guided learning mode in Gemini
Appwrite goes public (generally available)
Google’s Jules AI Agent goes public (generally available)

I think we can say August start week was a huge bag of announcements and model releases.

That’s it from this week! A fun exciting week, not at work, but having fun on the side.

See you in the next one!

Happy Coding :)

For more news, follow the Hackernewsletter (#757th edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #53

Sat, 02 Aug 2025 00:00:00 UTC

Week #53

A pretty slow moving week, but a lot of consumption. I watched and read a lot of resources on SQL, databases and backend in general. I want to understand SQL to learn what is the fundamental unit in CRUD apps, which I think is a SQL query.

Writing SQL query by hand in 2025 might be obsolete, but that 2% or 5% of the queries are pain in the ass to get LLMs write, that’s when the knowledge the depth, the legs of the PI shaped learning helps and make you feel alive, after weeks of existential crisis. I’ll keep learning SQL for the time being.

Admitting that I didn’t push any code to prod this week, it was a bit of struggle for me to get things in shape after loosing touch, I did a ton of debugging, but didn’t push any code. I’ll be honest about it, as there are days where you are just a helping hand or a gazing at the things unfold before you, because it’s ain’t my time or space to do it. Next week, probably will be breaking prod.

Quote of the week

“Without your involvement you can’t succeed. With your involvement you can’t fail.”
— APJ Abdul Kalam

Immerse yourself in the process, because you can’t win without, and the only thing in the way is yourself. I have involved myself in programming completely for 3 years, and have not failed in a literal sense, I have a great job, have good motivation to work towards, yes I made mistakes and continue to learn from them and make more different ones, that’s the process. If you put out, you know, you learn, you get better, the more you put out, the more luck doubles up each time.

Most people get exhausted after the upteenth attempt, but they might be just one more step away from success, success should never be the only goal, if it is, what after you achieve it? The process, the madness to do it, should be, the curiosity that will flicker in the toughest of times will carry the weakest to the peaks and the absence of it can make the strongest tumble down the cliff.

Created

Wrote
- TIL: TMUX Scroll mode select and copy
  - I was in dilemma, I sometimes use Ghostty or default terminal, sometimes with multiple windows, sometimes different tabs, sometimes zellij and sometimes tmux.
  - I just forced myself to use TMUX and ended on the problems or flows that I have been avoiding to fix,failing to find and navigate around like a wizard, but took a few minutes to get the mess out, and here we go.
  - This was just a few key strokes away namely, prefix + [ to enter the scroll mode, Ctrl + Space to enter selection mode, and Enter or Ctrl + J to copy to clipboard. Rock and rolling with log driven debugging.
- Thoughts: Kubernetes isn’t for you
  - I agree to this post, in certain situations. Especially for me if I am starting out in tech, I probably would not give advice to learn Kubernetes. Just get dirty with the normal deployment options, understand the pain points and then switch to K8s or others as needed.
  - This post lists down the points that people use Kubernetes just for bragging or showing that they know what they are doing, or just using it to show that they are cooking. But in reality if you are a small scale team, with couple of services, using Kubernetes makes no sense.
- Thoughts: Just fucking use kubernetes
  - I agreed to this post, too, as I said it depends. If you purpose is to learn Kubernetes, be a DevOps guy, be a SRE, or some cool tech guy, nerding, curios about it, JUST F-ING LEARN KUBERNETES.
  - This post made really good attempts to pull the trigger for me, I’ll quote waylon walker for a giga-chad quote
    “But my App is small, So is you ambition”
  - That just hits home, for me at least. If you think your app is a toy project, you are probably right and wrong, how do you know that if you haven’t deployed yet and people haven’t used it yet.

Read

Overthinking GIS
- This just triggered some neurons in my brains, some horses running. I got a bit of interest in exploring GIS data. Want to play with it, and find interesting details.
- This post highlighted the ways to use GIS data to get the usability of a land, basically which are fertile and usable for farming I think. But the technicality in which it was explained was clicking the right knobs at the right time with the perfect steps and images.
- Would be certainly writing a post in this style sometimes soon.
HTTP VS Websockets: The breakthrough moment that clicked
- Such a honest, humble and insightful post. HTTP as a delivery truck, Websockets as a telephone! Such a relatable example, this shows the author got it right and is able to connect it well.
- I can’t agree more to this tip
  Try building a tiny app with both. Make a little dashboard that gets real time updates on something simple like stock prices or server load. Build one with HTTP polling. Build one with WebSockets. You will feel the difference, not just see it.
- Back when I was a freshman in college, I tried doing this kind of thing for my world atlas chat app game, and failed badly, calling database for each message to send over the HTTP, what a idiot I was, but then, google searched about this and found plethora of articles about websockets and what not.
- While making that, I realised the pain points getting solved with websockets and everything just clicked.
More than Code
- Reading code in the age of AI, is gold and the only thing I think people spend most of the time, with other being vibe-debugging which might be fair less then vibe-reading.
- Reading code should become like instinct, just by gazing, you should be able to smell bad code and sniff the bugs out. This is not easy, it comes with practise and years of slog-debugging, first debugging the human code then go to vibe-debugging.
Learning GRPC Completely in Golang
- This explains what GRPC, protocol-buffers and the connecting technologies behind them in Golang.
- This gives a great overview of what and how the APIs are created.
Death by AI
- Woah! AI Overview mode in Google is terse at certain things. But for programming, I think it works a charm 95% of the times. yes there are pretty bad hallucinations too due to reddit and slop debates.
- Haven’t thought that it would mistaken a person with other person’s name and call it dead! Hillarious.
Why I do programming
For me, programming has always been more than a skill. It’s a way to explore, to tinker, and to satisfy curiosity.
- This is what programming is about, using curiosity to find the solutions to the problems. Curiosity is like a fire that keeps you warm in the winter.
Why GenAI Infrastructure feels backward
- I agree to this, this post can’t read this entire thing, but can understand the point of view
- Python and Javascript seems to be taking the forefront in the infrastructure side of things. All AI-Labs first launch SDKs and Packages for these two ecosystems, but never the others which are suited for them like Go or Rust.
- It’s time to change and make a difference in this revolution.
What are JSON Web tokens (JWT)
- Find me a better JWT tutorial then this, I’ll wait.
- Explained the problem then developed the intuition for the reason why JWTs exist. Very well explained, detailed and the diagram made it perfect for visualising the flow. It also mentioned the best practises, ticking all the boxes for a great article.
- Must read for beginner getting to understand JWT Authentication after learning Session based authentication.

Watched

Anatomy of a Request: A deep dive of a http request processing from the Backend side
- Woah! That is a ton of computation.
- On Client: Creating the payload, encryption (write copy), loading in kernel space, sending the data
- On backend: Received the data, reading to the user space, decryption, decoding (serialization) of the body.
- So many steps are there, the speaker rightly said, its a fascinating field, the more you go deeper, the more stuff is there to explore and learn.

Writing a Text Editor - Computerphile
- Interesting that text editors use Gap buffers. Its like a temporary register (block of memory) used to append text characters while the user edits (adds) to the file and then it gets saved, the remaining empty part is truncated. Clever data structure.
- He explained it so nicely, the approach, the problem and then a new intuition, again a limitation, then a proper intuition.
- Vi uses linked linked like data structure for editing, very interesting.

Why do databases store data in B+ Trees
- Everything is about intuition it seems, you see a problem, you think for a while, you think of a ideal scenario and you just scramble up a solution by adding the good parts and discarding the bad parts.
- This was the same, for why database use B+ Trees. Because we need to optimise for any arbitrary access for data as well as for range queries.

SQLite: How it works: Richard Hipp
- What a banger of a presentation and talk. Explained so much, in depth, in such a short time. It helped me understand what SQLite actually is, it’s a parser + virtual machine to run the core part and basically the fopen function in C to actually perform the operation.
- One unique insight here is
  - Reading 10 files content from disk is slower than reading those file contents from SQLite
  - Why? Because the database file is opened once and the reading happens in that instance only, data is stored in pages (fragments of memory), so it’s just a matter of reading bytes at a specific order.
  - But reading 10 different files on disk will make you use fopen 10 times, and that is slow!
  - 200 IQ move from SQLite team, have never seen such a beautiful solution to almost all the problems in the data world.

The real reason you can’t get a job
- More work, more luck
- Curiosity to learn, outperforms desperation to get money

Simple Joy of programming Course announcement
- It’s a great commitment to teaching the fundamentals.
- Not just fundamentals but building on top of the strong foundation.

Sync Engines and Local Data: Discussion of different database sync providers
- This actually made me a bit curious about sync engines further. I was in confusion when I heard about it from Theo as he used it for T3 chat. It didn’t made sense at that time. It still doesn’t, as why syncing is required in a chat app, all the data comes from the backend, there is no processing on the frontend?

Learnt

Getting into scroll mode in TMUX, selecting text and copying with vi-like keybindings.

You can’t do a unpacking in a sql query when using nested query for more than one column returned

Like example i have a query like this

SELECT 
  s.id, 
  s.package_id,
  address_id, 
  a.address AS from_address, 
  (
    SELECT 
      address,
    FROM 
      addresses 
    WHERE 
      id = p.to_address_id
  ) AS to_address 
FROM 
  scans AS s 
  INNER JOIN packages AS p ON p.id = s.package_id 
  INNER JOIN addresses AS a on a.id = p.from_address_id;

But let’s say for some reason I wanted to also get the to_address_type like a column from the addresses table

You might try to over-optimise the queries and try something like this

SELECT 
  s.id, 
  s.package_id, 
  action, 
  contents, 
  address_id, 
  a.address AS from_address, 
  (
    SELECT 
      address, 
      type 
    FROM 
      addresses 
    WHERE 
      id = p.to_address_id
  ) AS (to_address, to_address_type) 
FROM 
  scans AS s 
  INNER JOIN packages AS p ON p.id = s.package_id 
  INNER JOIN addresses AS a on a.id = p.from_address_id 
WHERE 
  address = '900 Somerville Avenue' 
  AND s.action = 'Drop';

And ERROR, you can’t do that

This bit right here

(SELECT address, type FROM addresses WHERE id=p.to_address_id)
AS
(to_address, to_address_type)

This is not feasible in SQL, you can’t unpack multiple columns from a subquery directly and alias them inline in a single SELECT clause.
Well I have to do it this way then, duhh

SELECT 
  s.id, 
  s.package_id, 
  action, 
  contents, 
  address_id, 
  a.address AS from_address, 
  (
    SELECT 
      address 
    FROM 
      addresses 
    WHERE 
      id = p.to_address_id
  ) AS to_address, 
  (
    SELECT 
      type 
    FROM 
      addresses 
    WHERE 
      id = p.to_address_id
  ) AS to_address_type 
FROM 
  scans AS s 
  INNER JOIN packages AS p ON p.id = s.package_id 
  INNER JOIN addresses AS a on a.id = p.from_address_id

What a long query!
By the way, this is one of the questions in CS50 SQL Problem set 1 packages section.
Any better way to do this? drop them in the comments or hit me up on my socials, will be completing more challenges this weekend.

Tech News

Google Gemini 2.5 Deep Thinking mode on app
- This looks something interesting, Google keeps pushing the boundaries for what LLMs are capable and making it more accessible.
NotebookLM launches Video overview mode for notebooks summaries
- This is a good feature, not great! I tried it and it’s decent, cuts in between, just text and arrows matched up like a presentation deck for the given problem. Not much value for a deeper dive, but really great to get a gist and overview.
- I can see this being used for news aggregation or summarising things from a lots of similar sources.
ZAI release GLM 4.5
- Another chinese model that is a serious contender for coding or general purpose tasks. That too local and open source models.
Qwen releases 3rd version of Instruct, Thinking and Coder models
- They launched quite a few variations, 235B for thinking and instruct
- Also 480 and 30 Billion variant for the Coder model
POE introduces the API for AI Models
- I have used POE in the past, and this looks like a great addition to test out a few SOTA models for free with some limitations via the API.

Phew! a lot of open source models are cooking! Everything is intuition and curiosity, and AI is still a thing to wrap our heads around, but the fundamentals are the same.

See you in the next one!

Happy Coding :)

For more news, follow the Hackernewsletter (#756th edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #52

Sat, 26 Jul 2025 00:00:00 UTC

Week #52

Finally a rewarding week, after 2 week of slogs. I / We were able to make the client happy after having gone through hundreds of issues and wired experiments. I was involved when the bug was introduced, but I had clarified that is this what we want, and at that time the answer was yes! Still in the end I was the one biting teeth and on the burn.

Anyways, at least the problem got solved and we were saved.

I read a ton about SQL and SQLite this week. Want to more this week too. I have a few ideas and issues that I am finding interest to work towards and finally at the end of the week or weekend will have something to share.

I am considering a mini series of SQL tib-bits, not a thoughtful things, but have something in mind over the month.

Ops! almost missed it, it’s week #52, that is, one year anniversary of Techstructive Weekly. I can’t believe that. I just kept writing and here we are with 50 weeks of constant reflection and hitting the publish button.

I am not going to stop this, instead I fancy myself writing more over the months and years, because some has said it.

Writing is thinking and thinking is writing.

There might be 104th edition of this newsletter with nothing new, maybe it will be completely changed, who knows what the future holds, but one thing is certain, something will change in me because of this. A hope, and not a false one.

Quote of the week

"A year from now you may wish you had started today."
— Karen Lamb

I don’t regret even a slightest that I started writing this weekly a year back. I learnt to articulate better, knew things I forget in a day or two. Helped me organise my learning and reading resources, be mindful about the media that I consume, helped me get over the blank page fear and most importantly helped me cultivate the reflective mindset rather than what next. Because this period when I write this is important for planing and thinking next what to do.

So, what is something I wish I would do now?

Log the things and observations about LLMs and different models or even versions. Different techniques about coding with LLMs that work and don’t work
Monthly Open source contribution reflections
Daily Work log (private google doc) inspired from Arpit’s ritual

Will I be able to do all? Probably not, but even something better at the end of the year.

Read

Here are some of the articles and tid-bits around the internet I read in this week:

SQLite Internals: Pages and B-Trees
1. This is quite interesting and helpful in making things clear
2. Every piece of data is stored in pages, a page is the unit of data in SQLite. Each page has parts like divided each for storing its metadata and the actual data.
3. Each type has certain number of bytes to be stored, so there is a identifier for that, so it makes retrieval and storing efficient.
How I keep up with AI progress and you should too
1. Simon Wilson, Andrej Karapathy and official sources are quite the right streams of sources for learning and keeping up with AI and tech in general.
2. These people are really hands-on and have a decades of experience and wisdom.
Rickrolling Turso
1. Avi has been hacking on Turso, he is a database nerd.
2. This was a fun way to demonstrate how to hack into a rust database, sqlite internals.
Why LLMs struggle with analytics, and how tinybird solve it
It ain’t about generating perfect queries, its about getting insights about data.
1. That’s a great way to put it.
2. With LLM context is really important and especially for critical things like database queries. Passing only the relevant info and cutting the fluff out is the key, but executing SQL queries in itself is that problem for the analyser.
3. So LLM is just solving that step but for writing the actual query from the natural language.
Simon Wilson’s commentary on GitHub Spark
1. As per this and more experimentation, it could be said Spark is the most competent and well engineered AI Assisted Code Environment.
2. Why? Because it’s GitHub, it has integrated the LLM in such a way that it seamlessly blends with GitHub Actions, Pages, Repositories, Secrets and Environments, Google had a great chance with FireStudio, but failed there.
3. GitHub seems to have taken a better and more engineered approach and not rushed it.
The Business case for Vanilla JS
1. Just write plain javascript and shut it up. Why are we really bottling ourselves with the gazillions of frameworks and mental load of understanding each other’s semantics and constraints and styles.
2. Why is there a need to create abstraction on top of abstractions. I get it, that writing javascript is cubersome each time. But the amount of abstraction and the cost of it adds is wild. We have a separate steps and ecosystem for doing just these, man!
Memcached Architecture
1. Simple and straightforward explanation of the memcache
2. Simple in-memory key-value store with slab-based memory management to avoid fragmentation. It has a threaded architecture and per-slab LRU for efficient concurrency and eviction, Also client-managed sharding enables distributed caching without server communication.
Rate Limiting explained in 252 words
how many requests a user or client can make in a given period
And not control of how much a user can consume over a longer period

Watched

Some random youtube video I watched at night after work during the week and weekends:

DHH on Lex Fridman Podcast
- So many hot takes and might change my perspective on software engineering
- Simplify development is important than elegance, eg. Ruby over Python, SQL over ORMs
- Using editor to chisel out code, and not let AI rip it.
- Finally someone to speak about the craft of coding, the fun of writing code alone, in the darkness, letting us spare time and let us cook.
- Engineering managers are meaningless for startups.

SQLite’s WAL Mode is faster than DELETE Mode
- This is so well explained, first showed everything what each one is and then the benchmark just makes everything clear.
- The WAL mode basically writes the changes in a separate file and merges to the original db file whenever required, hence there is no overhead when reading or writing multiple writers or readers.
- The delete mode is like a backup, a journal, it keeps pages of the data that are to be changed and after it is committed it deletes the file, that clearly looks slow.

Learnt

A few things I learnt while at my work and majorly side-questing after work:

DUAL Table in Oracle and MySQL Databases
- Dual is like a dummy table that you can reference in Oracle or MySQL databases as in these databases you cannot use query without a `FROM` clause.
- So `FROM` clause is requried, and hence the dummy table is needed to satisfy that quirky semantic of those databases.
SQL Sequence table
- The `sqlite_sequence` table is internal table that helps sqlite keep track of the auto increment counter for each table’s primary key.
- It has 2 columns, name containing the name of the table and `seq` denoting the max value of the id (auto-increment value), usually the last inserted id of the row in the table.
- Key things to note are:
  - Only one entry for each table, so this will only create the entry for the table if it has a integer primary key with auto-increment constraint.
  - If you change it manually, SQLite will still select the max of the row primary key by getting the full-scan on that table.

Tech News

Some of the things happening in tech, hot and smoking.

Qwen-3 Coder: It seems we have now a whole bunch of terminal based ai code war. Claude Code, Codex (does anyone use it?), Warp, Ampcode, Gemini CLI, now Qwen.
GitHub releases Spark for public preview: This looks solid and well engineered product, will be trying out this weekend, finger crossed, will get time and will be streaming it.
Google is testing an vibe coding app called Opal: Another one. We have no shortage of these, Firebase Studio, Lovable, V0, Bolt, and what nots.

I’ll leave you here, it’s been a great week atleast on the learning side, should be more fun next week and over the weekend.

For more news, follow the Hackernewsletter (#755th edition) , and for software development/coding articles, join daily.dev.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #51

Sat, 19 Jul 2025 00:00:00 UTC

Week #51

Another hectic week, the work seems unstoppable and almost killing my back. Had to work almost 12 hours at stretch to get something in production. Not the best of times to be working at. Especially when life puts you in wired position, not just in work but in mental situation, you have to believe in yourself and amidst that every other bit of things happen.

Random Thought

I was working away some bugs and fixing issues which was created for some other purpose at the time of initial prototype. But as the time evolved, it changed and deviated from the initial purpose, so fixed the current problem at hand. But people then point out your decisions. I felt grieved, lost, and most importantly wrong. But then things clarify, working all the possibilities of blunder out. Only problem was the decision made by the person in front of me when I asked the question in the past. But it was my fault in the end today. Not blaming anyone, just letting my thoughts out that people are too quick to judge and blame you when things go wrong rather than clarifying what they discussed first. World is cruel for the symphatic, for the humble, for the kind. It’s just is life. Can’t do much. But will you with one inspiration if you are lashed out at and the tides are against you.

Crying is never a sign of your grief, its a sign that you are alive right from your birth.

Is it okay to cry? Probably yes, but don’t feel guilt and shame in crying out, cry alone but wipe out the tears and move on. At the end its only you who have to wipe your own tears, there are people who are around you who care, but crying in front of them might make them sad that’s what you would think. I too think that and cry alone. The journey is lonely, there are people that will help you, guide you, love you but in the end you need to put your foot ahead and walk yourself.

Quote of the week

“The oak fought the wind and was broken, the willow bent when it must and survived.”
— Robert Jordan, The Fires of Heaven

Such a great quote to describe my situation right now. I am just bowing down, accepting the flaws but not the things that were not mine. Just letting go of things, and let life show its wrath or the good behind the bad its happening. Because I can confidently say, every storm ends with a new beginning and there are definitely some things that flourish after the storm but some do get weathered out due to resistance.

Read

How I do it - Daniel Stenberg(Creator and Maintainer of cURL)
1. Such an honest and encouraging post. It shows how a single curiosity, a fun project can become a livelihood for someone or many people.
2. He has made his curiosity take direction and kindled it with consistency.
3. He also without shame and guilt admits the truth about maintaining a project or even working at something for quite a few years or even half a decade. Commendable spirit.
  First let’s not pretend that it always feels fun and thrilling. Sometimes it actually feels a bit boring and done. There is no shame in that and it is not strange or odd. Such periods come and go. When they come, I might do less curl for a while. Or maybe find a corner of the project that is not important but could be fun to poke at. I have learned that these periods come and go.
Reflections on OpenAI
1. A good type of post actually. Very rarely I have seen people reflect back on the company they worked and detailed the things he learnt and felt at the company without brag or hype especially for a company like OpenAI
2. Everyone does a post where they talk about where they want to go next, or just complain about the things that made them leave the org. An important distinction for this post.
3. It seem fascinating how bold and achieved is OpenAI, the risk and rewards had been blessed to OpenAI it seems, I am not saying they have not worked hard, just that it could have gone worse, if ChatGPT didn’t pick up.
Kubernetes is not just for Black Fridays
1. I mean this is correct, I don’t want to deny this. Having a dependable system that scales is a good thing, instead of hacking a script or a pipeline for deployment or CI, creating a heavy yet robust system is worth it.
2. However, the only counterpoint I have is how you learn the things that the larger system does without using each component at its core.
The great SQLite rewrite
1. It is a serious thing.
2. People at Turso are seriously building it, it’s not a honey pot, last week I think they rewarded a thousand or more money to a person who actually found a bug in the Turso Rust version.

Watched

Philosophy is the last subject worth learning and studying amidst the AI hype
- This might look like off-topic, but it really is relatable and worth thinking about
- If you think, AI can do everything, what is there that makes human live?
- What makes humans, human? the communication, the art, the craft, the way to express ideas? IF so why are we in a race to get that replaced with AI
- So, he clearly distinguished the two things like theoretical and philosophical things. The former might be replaced by AI, but the later is the crux of what makes humans, human.
- Worth thinking and reflecting upon if you are getting anxiety as a person with the new LLMs launching every day.

Learnt

SQLC Issue with PostgreSQL Aggregate types
- I was just trying to get back into open-source, just to distract myself and gain the lost fun back in programming especially after hard times at work.
- I scampered through the issue list at sqlc and found this bug that caught me and wanted to fix.
- Reproduced the issue, gave the filestructure to GPT, Claude and Grok and it gave me a few files that I could look for and find the bug after log slog and print debugging.
- I reached a point after a few hours of interrupted debugging for a couple of days and found the core part where it misses the type.
- There is no type inference done for the return type of the aggregate function having multiple schema type, the table schema is not being passed to the function that does it, so it just picks up the first map.
Completed SQL Bolt course tutorial
- Completed all the 18 exercises of the learning SQL starter guide on SQL Bolt
- Learned a lot of interesting quirks about LEFT JOIN and RIGHT and NATURAL JOIN being only present with LEFT JOIN in the sqlite js with the wasm compatible in the browser.

Tech News

Kimi AI K2: This is probably the largest open weights model out in the wild, and that too probably the first that is really beating the state of the art models in tool calling and agentic coding. A huge standard leveller for open source community. There are quirks about the license but its pretty commendable in terms of what they have achieved.
Kiro IDE: Amazon out of thin air just blew this up. Its almost like Cursor but with more controlled and flexible using preferences. It almost forces developers to plan and thoroughly consider the most tedious and boring part (sometimes the most exciting) of the development which is defining the problem and listing the features.
Cognition acquires Windsurf: Devin will now use Windsurf, interesting deal this. How on earth are these two even partnering, doesn’t makes sense. At one point it looked they are rivals, even opposite mentality, but somehow they are collaborating now.
ChatGPT releases Agents: This is like a thing that we all knew was coming but is late on the tables. Need to check it out if that is available for free users or not. If yes, then will be quite a fun thing to do and learn.

Didn’t consume much this week, produced a lot of shitty python and probably fixed a tons of bugs, but never got recognition. That’s quite a whirlwind to do, will be taking this weekend a slight self development route to upskill myself for learning SQL. I am finding it really interesting to learn SQL and contribute to SQLC, will continue on that.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

For more news, follow the Hackernewsletter (#754 edition) , and for software development/coding articles, join daily.dev.

Techstructive Weekly #50

Sat, 12 Jul 2025 00:00:00 UTC

Week #50

A bit of boring week, had a lot of hopes starting on Monday, but it just tumbled down. Also broke the 15 day long writing streak. It all just falls apart, I want to change things now. I work remotely, and not able to balance work-life and personal life, its just too exhausting. I get self-doubt creep up with this, am I really a sustainable developer, can I not continue development for the next 10 years even? It is starting to haunt me. I know it would be temporary, imposter syndrome feeling, but need to really change the flow of my daily routine to get a mindset shift.

But learnt a lot this week, about SQL continuing from the past week, started solving SQL 50 problems, which I am halfway through. Also need to ramp up my problem solving skills, need to build projects, there is lot of stuff to do. I just get overwhelmed with stuff every time.

I haven’t streamed for the past 3 weeks and it is giving me negative cloud of procrastination. Hopefully will be able to break the jinx this weekend. July has been good to me in the past, hoping for the best again.

Quote of the week

“It’s not about how hard you hit. It’s about how hard you can get hit and keep moving forward.”
– Rocky Balboa

Yes, I admit I might be hit, I have fallen, I no longer have the motivation to code for some patches of the week. But I have the spark back at times and flash of seconds. It might be a temporary burst of problems at my end, but every storm weathers out, every night ends and that’s when the comeback begins, the sun strikes and light appears, there we see a hope, and the light at the end of the tunnel, it might be small, but enough to keep moving us ahead.

I am willing to change my workflow and routine, I just need to give it some time. If habits are easy to build they are easy to break too. I know this with my 15 day writing streak that I broke, I didn’t had the mental energy to sit and write for an hour, and that’s the reason I am frustrated with my life, with my work. I am adamant to change.

By the end of July the max, the first year anniversary of this newsletter will be special one.

Created

SqueaLite Browser Editor → squealite.vercel.app
- Just wanted a quick and easy way to write SQL queries and test out on a database.
- Turns out sql.js is a great solution for it, it’s a WASM library to interact with SQL in the browser with JS. I am using it to create a database for that session for the user in the browser and let the user use it as a playground database.
- I added (AI-assisted) editor, result panel, multiple query tabs, loading and saving queries and databases, among other visual aided table things.
- Planning to add two more sections
  - Problems → Like leetcode styled run and submit challenges.
  - Exercises → Independent and specific playground for understanding specific concepts.
  - Per user database for playground and exercises using Turso libsql database.
Analogies are not always good: Comparing VACUUM in Postgres with free in C
- Well, this is a off-topic write-up I felt writing because no one seems to have compared VACUUM in Postgres and free in C.
- Those 2 might look unrelated but for understanding it makes a lot of sense. Actually I learnt that VACUUM is not like free, but VACUUM FULL does look like free.
- My point is drawing analogies is good but not when you are not sure of either of them, it could look good on the high level talks, but getting a little deeper will shatter the comparison if not thought carefully.

Read

Tools: Code is all you need
- MCP is not worth the context I think, and this proves it. I have no experience but just from the understanding it would make sense, since a lot of things need to be made clear before using them right?
- Imagine a human navigating this, you will have to make him aware that these are the tools, explain what each tool does and then provide them. But not to a professional, so we can conclude that LLMs are not professionals yet!
- MCP is a bit verbose and might not be right for all kinds of models with less context window or limited capabilities.
Thoughts on motivation and my 40 year-old career
- This was a intense read, a bit of self reflection and some insights into business vs wage job.
- I think engineering is hard to explain by the looks of the post, its a bit of complex thing to be working at, not hard, but you need time and patience to get used to it.
I still care about the code
- This is so well put, LLMs are not compilers or interpreters, they still are unreliable. I mean ~~software is unreliable~~, untested software is unreliable but LLM generated code is on the extreme end of black holes.
- Hallucination is a things, I am not talking about on-surface wrong things, I am talking about deep and little details, that right now only human developers can craft.
How AI is changing the Software Hiring in 2025
- Being low on confidence and not fake ego are so valuable, being humble has never been rarer and rewarding then ever.
Waiting for a Follow-up
- Man, this week, I am reading and watching stuff, almost as I do them, it was the analogy thing, now I am waiting for a follow-up, it ended but still, feeling like I am walking into coincidences right and left this week.
- But true, looking outward helps. It creates a sense of motion and drive us from the overthinking and procrastination that would have consumed and stalled us.
How Cloudflare’s R2 actually works
- It now makes sense that how R2 is ideal for storing and frequently accessing large amounts of binary or media like data.
How to get a job at Canonical
- Care, Attention to detail and skills are quite the standard in any big tech company.
- Interesting and exciting to see Canonical not using AI, at least there is hope.

Watched

Learning Pydantic AI in Marimo
- This was the video of the week, I learnt a lot of stuff. One being, marimo is something I need to try as fast as possible and make it my go to tool for anything prototyping in python.
- The discussion around Agent and LLMs is so beautiful.
- Adding Type_1 | str is so so useful, It just opens up so many possibilities. The entire Agent concept could be made possible due to this.
- Because if you are expecting a type in return but there is no enough information to return, the model has to forcefully return a value which might not be great, by just adding a | str it would then ask the user to fill in the details. Extremely well designed library.

“Software is changing again by Anderj Karapathy” Review by the Primeagen
- Analogies man, they are everywhere.
- Andrej drew a few analogies like OS and Electricity with AI and LLMs, which makes sense but Primeagen rightly said that he could come up with any thing and compare it with anything if he has too, but understanding deeper just breaks the analogies, which was the reason I was writing the before mentioned post, all of this clicked to me.

Why Programs use Stack, Heap and other memory segments
- Now, I got a better understanding of what heap and stack actually means.
- Stack would be useful for quickly adding function calls, and heap for accessing long term variables on the memory which might be fragmented.

Learnt

SQL
- SQLite doesn’t have TRUNCATE
- .read is a handy way of reading sql queries in a shell environment
- CASE WHEN <CONDITION> THEN <value> ELSE <value> END
  - This is the statement I used to filter values in a aggregate functions like COUNT, SUM, etc.
  - Really helpful to know this as it could be used with certain conditions to count specific elements but not for the entire query. Or there are different conditions for each column to get out.
PostgreSQL
- Timestamp and Datetime
  - Use named timezones as far as possible and stick to UTC
  - If you are using both date and timestamp better use datetime
- UUID
  - Are efficient compared to text type
- Integers | Numeric | Floating point
  - Integers are fast, range from 2, 4 and 8 bytes, super accurate and fast
  - Numeric > extremely precise but slower
  - Floating point > fast with a little less precision
- Don’t use money type when dealing with money

Tech News

XAI releases Grok 4: The model is the smartest model to date, and also the most expensive. Interesting quirk is that, if you ask for its opinions, it dials up Elon Musk’s Tweets to find his opinion. Wired quirk but maybe *unintentional*
Cursor Pricing drama: Cursor just updated / clarified their pricing for the individual plan and being costing a bit high for unlimited usage.
Mistral updates the Devstral (Coding LLM) small and medium variants: This makes the Local Coding model race a bit heated, as Devstral Small stands neck to neck with Gemini and GPT 4.1 models.
Perplexity launches Comet browser: Now AI is getting into everything. I am not sure when this will end. People are just too eager to cram AI in the silliest of things.

As expected, there was a model release this week, if you have been following the trend, we are nearly getting a new model every two weeks. And for the past two weeks, there was no major model drop and this was anticipated. Next week Open AI might be cooking something.

That’s it from the week 50, we are close on the one year anniversary of this newsletter. Wow! a whole year has been gone in writing these newsletters, didn’t expect it to be that quick.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

For more news, follow the Hackernewsletter (#753 edition) , and for software development/coding articles, join daily.dev.

Techstructive Weekly #49

Sat, 05 Jul 2025 00:00:00 UTC

Week #49

I’ll keep this week short, I have a lot to learn, and too little time. I have been studying SQL in detail, practising problem sets over the week. Taking a few more courses on Redis, Python, and more SQL.

One interesting thing, I kept on with my creative writing endeavour and made 14 days streak, writing with an average of 1200 words a day. So, I am roughly at 20k words after 15 days on my novel. This could be my first novel, could be even ready by the end of July if I keep going after it. Sounds exciting to me. Can’t wait to complete it.

I also want to create a weekly wrap-up and learning for SQL. I am really ready to dive deeper into the database world. I think this is what I can master and become a depth-first person for Databases and SQL.

Quote of the week

“Every now and then a man's mind is stretched by a new idea or sensation, and never shrinks back to its former dimensions.”
— Oliver Wendell Holmes, Sr.

I kept on writing my novel, one hour each day, I had no idea about the whole story, just kept a timer and no distractions, ideas flew after a while and the word count reached 1k, that momentum is enough to carry it ahead and continue the story next day and the day after, just like the mind has been stretched and formed with new ideas and uncharted water. Here imagination and curiosity take control and it feels so empowering and refreshing to discover that something existed within but never tapped.

Keep your mind open to new ideas, experiment, and you never know something will click like a light bulb and could change your life forever.

Read

Which LLM writes the best analytical SQL
- This was a great tutorial. It highlighted the point that “SQL is still a skill”. LLMs can generate SQL, but not analytical and efficient queries yet.
- The cost is a factor, but right now it seems not worth it.
- I thought LLMs were replacing people writing SQL, but here we are still requiring the domain experts. The people who will roll up the sleeve and press CAPs Lock and head to write SQL.
- GPT models are decent in a balance of latency with accuracy, then comes Claude with high accuracy but slight slow, Gemini is good, especially the Pro, but takes time, the flash models are bad with faster times.
- So, the LLMs are not perfect yet, they might get a few years time down the road, but it looks far from now.
How long context fails
- This was an insightful observation, must know things before working with LLMs having a large context window, or even a short for that matter.
- The needle in the haystack problem is not solved, and LLMs are very sensitive to getting up in a rabbit hole.
  - Context Poisoning
  - Context Distraction
  - Context Confusion
  - Context Clash
- All of these reasons look the same, but can mean different things in different “CONTEXT”.
How we accidentally solved robotics by watching 1M hours of YouTube
- This is hilariously funny.
- They trained the model on a corpus of YouTube videos, because sometimes, text and images aren’t enough for these kinds of operations like robotics, where moments are essential for learning and iteration
Claude Code is my Computer
- I see this with many developers, not just newbies or just soydevs, but also veteran and knowledgeable developers, too. They seem to have a liking towards Claude code.
- I know the feeling I think after using Amp, Warp, Gemini CLI to some extent.
- But this is a bit of avoiding the chores part, which is fine.
Just fucking use Kubernetes
- This article is true and motivating to learn Kubernetes
- I read an article which was the opposite of this, not in a bad way, just a way of presenting the idea to keep it simple and avoid the complexity if not needed.
- This article, however, is a motivator in a couple of sense, first it says if you think the scale is low, your ambitions are low. What a statement.
- And also it’s an under kill if you think Kubernetes is an overkill
- Nice AI-generated slop, actually is good.
Writing code was never the bottleneck
- Yes, code was never the bottleneck, but it was not also the easiest thing. People plan and plan, and plan more, but the execution is not up the mark, and the planning was of no use. Coding is neglected among managers but empathy is needed to bridge the gap.
Why I want to write again
- I also want to write again. It has been almost a year since I started this newsletter, and I have not been able to break into a habit of writing the blogs that I used to back then.
- The author’s claims are so true, and I can’t agree more, but the habit is the key part that holds me back. There is never a smooth life; something or the other hinders the focus.

Watched

Understanding B Trees: The data structure behind databases
- This was super cool, the way it was visualized and taught.
- I adore Brain Yu, he is a master teacher. I learnt Python and Django from him.
- B-trees are basically trees with an equal number of leaf nodes. No unbalanced roots. The operations to make the insertion and deletion made it really awesome for understanding.

Agentic Coding: The future of software development with agents
- See, again, Claude's code and agentic tools are getting adorned by experienced developers too. This is the thing that makes me concerned about my own opinion and thoughts.
- It surely is powerful but it hasn’t clicked for me yet I think.

Vercel Finally Caught up
- Vercel now makes you pay for only the CPU usage and not the time.
- Still expensive from Cloudflare but a good dip in the number

I finally switched to PostgreSQL
- Planetscale just added PostgreSQL as a supported database
- Convex now switched to PostgreSQL
- So, convex becomes a free tier for Postgres on Planetscale

CS50 SQL Lectures Playlist
- I am learning SQL with this playlist, its in depth and also has a problem set to explore on our own.
- I have completed watching all the videos and just yesterday completed problem set 1, will be moving on to solving more.

Learnt

SQL
- Nested Queries
  - With this we can use select and that result set becomes a value for a condition for the outer select that is handy
  - = for single value and IN for range of values(list)
- JOINS
  - INNER JOINS are only for matching records between two tables
  - LEFT JOIN will join everything on the left (first) table, even if there is no matching record on the right
  - RIGHT JOIN will join everything on the right (second) table even if there is no matching record on the left
  - FULL JOIN will join both tables, populating the NULL values if either of them doesn’t match.
  - NATURAL JOIN is the most unnatural part of joins which will combine the tables with the common named column(s) and remove the duplicate column
- VIEWS
  - This is so cool. We can basically create this for reducing redundancy, Its like a macro, a stored function kind of thing, where it doesn’t have memory of its own but has things that it can run on existing data. Super useful to know

Tech News

A week a bit of no new models, nothing fancy launches. Finally, we get a week or two before it gets wild. It’s time to reflect and build something meaningful. It’s time to learn and plant the seeds that will help in the future.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public, so feel free to share it.

For more news, follow the Hackernewsletter (#753 edition) , and for software development/coding articles, join daily.dev.

Techstructive Weekly #48

Sat, 28 Jun 2025 00:00:00 UTC

Week #48

An fruitful week, not much on tech created, but a lot of learning and revising about the first principal thinking.

I tried to create a Pocket clone in Typescript with some AI-assisted (a bit of vibe coding) because I don’t like writing frontend. I failed a few times, and even my network seems to be off for a while, so I had to scrape that idea and stop streaming for the weekend. So, it didn’t quite work out. Instead this weekend I’ll only stream on Saturday to start one project.

I read and consumed a lot of stuff this week. I need to be creating and writing more too. The balance is essential.

I will share one small achievement and progress on my non-tech journey. I wrote 8,000 words approximately on one my Novel (or novella I don’t know yet). I wrote consistently for a week, 1 hour every day, averaging of 1200 words per day. A good start I think, better late then never. I have been procrastinating this for quite some time, in terms of years now. Finally Friday, 20th June 2025 was the day I beat that devil and my hear finally won. It looks small, on the out, but I think this is a start of something big. A turn that I want to remember when I reflect back on my life.

Quote of the week

“The thing that you are most resisting is probably the thing you most need to do.”
— Steven Pressfield, The War of Art

I have read the book, the war of art, and this quote speaks to me like fire. Something has resisted me from getting into fantasy and creative writing. But here I am after a week of writing consistently. I feel liberated and joyed. There is so much to do, too little time, too many pages to fill, too many characters to paint, too many stories to tell and most importantly so many lives to be kindled with writing.

Created

LLM plugin for code agent with Code toolbox
Tried to create a reader app for collecting and reading articles or content
- Like Mozilla’s Pocket, but focused on reading individual as well collected resources in a simplified interface, distraction free reading
- I get a lot of articles to read, sometimes, I get lost in between tabs and to be read articles, so I wanted to create a simple that lets me dump links to it and later I can skim through them for reading with a simple interface for reading.

Read

Pheonix. new Fly’s entry for AI coding agents
- This is a neat little thing from fly.io, they have made everything so right, that these AI Agents just fit perfectly in their ecosystem.
Using AI right now: A quick guide
- Gemini, ChatGPT and Claude these are the most widely adopted LLMs which are quite general purpose. But they also have nuanced quirks and preferences for certain tasks.
- Deep Research is a big deal and it has helped me learn a new concept every day. I use Grok AI, Gemini AI, Chat GPT and even mistral ai for specific things.
- Hallucinations are no longer a big deal in AI, they have reduced considerably over the years. It is still a problem but not what it used to be.
SQL Join flavors
- This was a insightful post, it helped me feed my hunger for learning more about SQL.
- The qualified, natural (Actually unnatural) and cross broad range of joins really helped understand the concepts better.
- The interactive examples are really great for those types of complex yet powerful concepts.
AI-free Writing
- Anton won’t use AI for writing, I am on his side. I write this newsletter and any writing for that matter without GPT or AI. Yes, I use it as a critique and help me understand how it reads, I read its thoughts and think and leave them. Next time, I be a little considerate about those mistakes.
Golang 1.25 interactive tour
- The json v2 package and the in-depth explanations are helpful in quick understanding of the changes. It has given me a good view on what is changing and is enough motivation for me to tinker on my own and create other examples.
Anatomy of a SQL Engine: Dolthub Blog
- This was a interesting post. It made me curious how a sql statement query is executed, the ast construction, then right recursive and left recursive. This blog actually made me take a in-depth guide on SQL.
- I have started reading about SQL and solving leetcode problems and even some problems on SQLBolt. I have even picked up reading “Learning SQL” Book for learning more. I want to get into “Database Internals”, that book is about how under-the-hood sql and databases work, for that I need to learn what SQL is in the first place.
Docker launches hardened base images
- DHI are the new thing in docker, they help in avoiding the image bloat.
- They are not just trimmed-down versions of existing containers, they’re built from the ground up which is really interesting to work with and help optimise the build process as well as deployment constraints.
Docker Hardened Images start with a dramatically reduced attack surface, up to 95% smaller, to limit exposure from the outset.
Kubernetes isn’t for you
- Kubernetes is made for the scale of google, where they have millions, billions of request per minute. Not for your 100 user startup. True, if the scale keep growing, or you have micro-service architecture, but if you have a bare-bones project and not many user-base, then sticking to simple traditional deployment is good enough
- I hate when people use shiny things for making themselves look like they know what they are doing, the soy devs, the gigachad 100x developers, and they only know what a pod actually means in a cluster.
- Why make things complex, it feels great at first for our egoistic brain, but believe me the complexity will hurt you in the long run. Maybe it will save you too if you are actually growing at scale. But the trade-off you need to decide.
Gemini CLI breakdown by Simon Wilson
- Google has really taken the time to build something of high-enough standard in the space of AI Agents in the terminal. This is a serious competition to Claude Code.
- I wonder what OpenAI missed with Codex. It just is bad. Not even usable to free users, which Google just took it by storm.
- The system prompt is really insightful, the tools are nicely laid out, its so simple yet well architectured.
Now might be the best time to learn software development
- The best time to learn software development was 5 years ago, now is the next best time to learn it.

Watched

How did they get their first Jobs
- Teej had some serious advice, having cover letters, showing that you care is so valuable. I have started to do that (oops, I am looking for a job actively, trying to get a switch after a year and half at my current company)
- Prime had a good take, it just happened to him after he had worked and seriously worked hard enough. He was just doing his best and suddenly luck gave him a chance and he took it with both hands.
- The other person, forgot the name. Also had a symphatical journey.
Cloudflare Containers
- Containers they are really a game changer. Cloudflare does things and does it on scale. This hits different. This is some serious stuff, maybe even a revolution in how we use and build software. It literally can spin up a new computer on the cloud.

Sam Altman on AGI, GPT-5 and other stuff on the Open AI Podcast
- This is all nice and good on the talk, they really need to answer why the heck are they even developing the models and selling them.
- I think the bubble is slowly going to burst and we are going to use AI surely but not the way we are thinking right now. Just as a tool.
- Maybe search will be revolutionised with AI, but I don’t see it in other places, it’s not worth the time and money.

Gemini CLI and comparison with other Agentic Terminal Agents
- Gemini CLI is really nice, at least for a free tier. I can use it and understand the LLMs capabilities. It helps junior developers get a taste of the software AI can produce, but these can really get double edged sword. Juniors will only produce AI slop and never learn anything.
- But yes, google is trying hard to make developers happy for now.

Learnt

Javascript has quirky behaviour with null, undefined and what equality operator
- We don’t know if the value is null or undefined, the object might be still undefined but it will be a truthy value
- The equality operator is very wired, this teaches us that too much flexibility is also bad
  - For instance, the ‘5’ == 5 will be true but ‘5’ === 5 won’t be
  - The triple equal is a type check whereas the double equal is a value check after type casting, which might be a bit unpredictable as the data gets complex.
Getting the unique elements from a list in Python without changing the order
- After python 3.7 changes the dictionary, the order of the keys inserted is maintained, so we can use that to create a map of the elements in the list as a key in the dictionary and return the unique elements.
- Neat little trick, could be well often be used widely in many cases.
```
# this might change the order
list(set(queue))

# this will preserve the order
# works for python > 3.7 
list(dict.fromkeys(queue))
```
Reference: Python Coding Stack: I want to remove duplicates from a python list

The Python Coding Stack

I Want to Remove Duplicates from a Python List • How Do I Do It?

Another short article today to figure out ways to remove duplicate values from a list. The ideal solution depends on what you really need…

7 months ago · 18 likes · 6 comments · Stephen Gruppetta

Tech News

Cloudflare Launches Containers
PNG is back with a new standard for APNG (gif like format)
Google drops Gemini CLI as a terminal AI Agent
Anthropic allows build, host, and share interactive AI-powered apps directly in the Claude app

There are lot of things happening for tooling around these LLMs, no model launches hopefully for a few weeks. But the launches from Google are just not stopping, constant in the news. Anthropic is now trying to expand in different directions after anchoring their flagship model.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

For more news, follow the Hackernewsletter (#752 edition) , and for software development/coding articles, join daily.dev.

Techstructive Weekly #47

Sat, 21 Jun 2025 00:00:00 UTC

Week #47

It was a exciting and exhaustive week, in a good way. I was able to almost complete a working AI Chatbot Application, start writing routine for an hour, get inspiration for more project ideas, and a lot of learning in tid-bits around LLMs.

This can’t be more exciting. A week or two before I was drenched in anxiety and negativity around LLMs as replacement for developers, but that’s far from reality and considering them as tools sounds so fun. I feel excited so much so that I feel I have no time to squeeze the energy.

This weekend I was not able to livestream and record any videos due to network issues and also I had a bit of cold. This week hopefully will be doing 2 3-hour live streams on creating projects.

Mozilla Firefox’s Pocket clone
LiteLLM Implementation in Golang

Not both of them, either of them will be working this weekend.

I haven’t created any videos, haven’t planned any video as of yet, but have a lot of ideas for writing. I think I will be creating a timer based blog writing, I have to force myself to write a blog. I have lost too many months writing a blog, can’t lose them anymore.

Quote of the week

“AI is thinking faster, so we need to think slow”
— I can’t find who quoted this, but I read it somewhere

This is such a great quote, whoever quoted it, awesome job. AI is producing slop and hit sometimes, it might not be correct every time, since perfection is the enemy of progress. So is math, thinking a problem for LLMs currently, they think a lot, hard, too hard at times, and spiral into overthinking just like humans, but meaningless and not required from machines.

We humans are so funny, we expect LLMs to work perfectly but have trained on the dataset that is imperfect.

We need to sift through the LLMs response, which could be slop or a gem, which ever it is, we need to think a bit slow than what it gave. Reflection is key in the case of AI-assisted anything.

Created

Bring Your Own Keys AI Chat App (BYOK AI Chat)
- GitHub Link: https://github.com/Mr-Destructive/byok-ai-chat
- Vibe Coded the Frontend with Lovable / Windsurf
- Half baked backend API with Claude and Grok AI
- Minor Improvements and Bug Fixes with Jules Agent
- Backend in Python (FastAPI), PostgreSQL Database
- Frontend with React
- Vercel + Render Deployment
- LiteLLM for model provider
- Yet to add proper authentication (probably with Clerk)

Read

The art of saying no: This post is so accurate in describing the feeling while doing AI assisted development. We have to constantly sift through hundreds of suggestions or ideas to get the one that we are looking for, and that is not easy and comforting at first. However, just like debugging intuitively, accepting suggestions and vibing would become a intuition based activity, it just takes a few misses and hits, some trials and explorations of how these LLMs work.
Never Just: This is absolutely relatable, when we are in complex problem solving situation, and someone says “why don’t we just make it simple”. The person saying this either doesn’t know the problem or knows too much that he simply can do it instead of asking it to others. If this just comes from the surface, it feels a bit hurting, it should be reframed to make it more informative and actionable instead of attacking.
A Linear Algebra Trick for computing Fibonacci Numbers Fast: An interesting way to compute Fibonacci numbers. A nice matrix multiplication trick.
A tale of two Claudes: A completely honest and realistic take.
1. Claude cannot work with Tailwind 4
2. Claude works well with compiler and gnarly debugging memory related issues
3. This is really cool to be aware of, we can use LLMs to guide us in the situations when we are not sure. Because it will speed things up and probably reach conclusions much faster and we can then decide if that conclusion was right or not, rather than spending hours in the gnarly bugs.
4. Someone said it, LLMs are thinking too fast, so we should think slow, what a perfect sentence.
Feedback is not attack: Feedback is quite a huge subject and can affect one’s relation in positive or negative way. Its not just about giving feedback, its also about empathy and being able to receive and accept the feedback well.
What would Kubernetes 2.0 be like: YAML should be replaced with HCL, that is a golden point. No like seriously, reading a long yaml file just gives me headache. I prefer JSON Instead. Also the helm package manager is written in Golang, I want to dive deeper into K8s.
This AI Agent should have been a SQL Query: This one was more like Apache Flink comparison and walkthrough, maybe as a example but it was good.

Watched

The State of Authentication
- This is so messed up. There are actually three parts in Authentication. I thought Authentication and Authorization, that’s it
- Authentication is basically “Am I who I say I am?”
- Authorisation is what I am allowed to do (if i am who I say I am)\
- Now the third part of Authentication
- Auth UI: Lo behold, UIs are not my thing, and yes Auth0 and some other probably some other services too.

Andrej Karapathy: Software is Changing
- We are in Software 3.0 Era
- Software 1.0 was code | Software 2.0 was Neural Networks
- Now we are in the LLM space, were we can send prompts to neural networks that can generate programs, that is wild idea
- We need to create systems around LLMs, like interface to interact LLMs, He puts it like we are in the 1960s era of Computing. Where the computers were not personal, they were mainframe, large machines that fit on factories.
- However the LLMs today are like those big gigantic computers that live in the cloud running in multiple clusters of GPUs, though we have local LLMs, the quality and accuracy of high parameter models is not near the local small sized models (they are improving). So we are reliant on Cloud inference like ChatGPT, Claude, Mistral, LLaMA, Gemini, etc.
- He relates the transition of LLMs with the augmentation to an autonomous agent. This is compared to Iron Man’s suite, which is a mixture of human control along with AI assistance. When he is outside the augmented mode, its his instinct that help him navigate, however, in the agent mode, the AI decides the track.
- Also the vibe coding term was emphasised which has a Wikipedia entry. Vibe coding is easy to do because we don’t care about the code, we care about the outcome, we can call it product driven prompting. And it makes sense when he said, we need to increase the Human-verification and AI-generation loop. It doesn’t mean, you’ll tell AI to one-shot the entire thing, instead go bit by bit, one thing and one action and feedback at a time.

99% of the AI Startups will be Dead by 2026
- Obivious, Startup is about survival and if all of them are wrappers around just a LLM they might die. But only if they are just that and nothing else. Turns out most startups might just be that.
- However, many unicorns and valuable startups are built due to trust and alignment of the problem solved by the product and faced by their users, so its about competition there.
- AI is here to stay and evolve, if the product keeps adapting, it will survive no matter what, the trend is, if people like it, people will pay to use it for comfort.

Why Internet went down for 2.5 hours on 12th-13th June 2025 (IST Timings)
- Half of the internet was down for almost 2.5 hours, due to Google Cloud outage.
- Cloudflare was down due to one service of theirs was reliant on Google Cloud, and that spiralled to all other services. Whoa! dependency hell is real.

Learnt

LLaMA actaully stands for Large Language Model Meta AI
A wired random prompt that surprisingly is not random: Tweet
- This is not 100% reproducible but quite the gist of it is similar
- Pick a number between 1 and 100
  - ChatGPT, LLaMA, Claude, Mistral gives 27
  - Grok AI gives 42
- Pick a random letter between a and z
  - ChatGPT, Mistral gives M or K
  - Grok AI, LLaMA gives K
- This is a wired behaviour from LLMs, and it also makes sense since it will try to mimic what it has seen most of the times in its training process.
Render is the new Heroku for free application deployment
- I used it to deploy my FastAPI backend for BYOK AI Chat App.
- Its similar to the old Heroku hobby tier plan, which was vindicated in 2022.
- I am not sure how reliable these Platform-As-A-Service are anymore after the Heroku sunset of hobby plan, but nonetheless its what we get from a free tier and its worth appreciating such companies are providing free tiers.
- I am wondering can Appwrite take over its spot, its more of a microservice oriented setup but still it has a lot of potential to compete in the space.
Copy and Paste Markdown in Google Docs
- This was a neat trick to learn, especially if you are writing for documentation and personal notes. Also handy if you want to quickly add something special without fluff in the doc. The keyboard centric approach helps in this markdown style.

Tech News

Duolingo launches Chess as a course
It wasn’t built by a huge team or chess experts. It started with a PM, a designer, and Cursor, an AI coding assistant. No engineers. No dev team.
Google Cloud Outage on 12-13th June 2025 for 2.5 hours
Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task

For more news, follow the Hackernewsletter (#751 edition) not arrived yet, and for software development/coding articles, join daily.dev.

Leave a comment

That’s it for this week. Hoped for a good week, got one, hoping for a better week.

Thanks for reading :)

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #46

Sat, 14 Jun 2025 00:00:00 UTC

Week #46

Life takes turns pretty quickly. I was here last week, complaining about life and here I am today, feeling grateful for whatever happened to me.

I took decisions last week and by mid-week, I was seeing results, like actual results. And I was just thrilled to experience it, when you decide to change, the universe listens and blesses you with opportunities. I am grateful for the things blessed to me through the week. Even if I fail next week, I have a head start to keep going and break the shackles one day.

Throughout the week, I dabbled in LLMs. Learning about different providers and models.

I spent 20 minutes playing this game before writing this newsletter. I was reading the Hacker News newsletter, and I saw this link and wow, the game is so cool. A binary search to find a word. A wordle but not quite that.

Quote of the week

“Practice doesn’t make perfect. Practice reduces the imperfection.”
— Toba Beta

I have been in the craft of programming for over 5 years, and now I know that it’s about figuring things out. I was naive and all over the place when I started, but slowly I learnt from mistakes and started to feel the patterns in solving problems. I am not perfect, never will be, but will have less muddier and novice like traits in me.

That’s what it means to live a developer’s life, really, adopting and adopting till your hunger and curiosity to know things is satisfied, it will be on you to keep it on till your last breathe.

Created

Bring your own keys (BYOK) AI Chat App
- Created entirely from Claude and Lovable AI
- Added Litellm for providers and model selection
- FastAPI for backend, Bun/React for Frontend
- Thinking of using a per-user database with Turso
LiteLLM listing providers and models
- This is a rough draft version of the post.
- I have worked together the way to list down the providers and models
- Looking at what else could be done and exploring different ways to query info about a specific model

Read

How I finally understood Docker and Kubernetes
- This was my pick of the week. I understood the reason why Kubernetes exists
- I knew the concept of Docker (it could be because I have used it extensively in the past to create projects as well in my internships to deploy APIs and apps)
- But the concept of Kubernetes is like a black-box. But the author’s explanation style and simple example made it clear.
- Kubernetes is like
  - Container Image > Deployment > Pod > Service
- Container Image is the actual image of your app that you want to run, maybe it has multiple of those.
- Deployment is like defining what and how many (other things too) to run.
- Pod is like the actual unit of containers; in itself, it has no control, it just runs whatever was given to it.
- Service is like the layer that exposes it to the world, maybe the network, the other containers, which are like a configurable exposure of the network.
I am disappointed in AI discourse
- Wow, at least someone feels like me. This was so relatable to read on a Sunday afternoon.
- This feels relatable to feel like how can people draw conclusions and biases when the technology is evolving everyday.
- People are either fully bullish on AI, i.e. by 2026, no developers needed, to some saying AI is still crap. Both of them don’t know what AI is capable yet. It is not that bad, compared to a year ago, it’s really good at generating almost accurate code, but not quite the touch that it can be untouched without a developer glancing and ficing it.
The Software Engineering Identity Crisis
- The quote that sums this well is this:
  Perhaps the most valuable skill in this new landscape isn’t prompt engineering or systems architecture, but adaptability - the willingness to evolve, to learn new skills, and to find your unique place in a rapidly changing field.
- Adoption is a skill that you can hone and be an engineer of the AI revolution
- The author thinks that there will be engineering in terms of managing and overlooking Agents or AI Systems. Also a good point mentioned is “The scope of engineering is not shrinking, it’s expanding”. I can see this point coming true in some sense as the level of understanding to drive these AI systems is better suitable to programmers and engineers. Since they already deal with the pain of solving problems. AI is no different.
- The first part that hits me the most is the loss of joy. I talked about it in the previous weeks. I don’t want to repeat that rambling, but yes, that somewhat feels a bit uncomfortable to digest.
- An excellent post in navigating this AI landscape while maintaining the core feelings of a developer and where it breaks.
Programming with Agents: Sketch.dev
- Great post on agents, good and expected take on agent definition
- David also narrates the need for agents and why they are suddenly in the hype. The function calling thing just blew the AI hype and at that time, the LLMs were not ready or trained for it. But in 2025, those LLMs are optimised for it, making it a great ecosystem to work towards.
- This actually raises questions: Do you really need the craft? The IDE? If the agent can do it for you at the speed of a prompt? Yes, LLMs are not gods, they need assistance, and when they go berserk, its on the developer to hone on his tools and hack it out of the mess.
The Gentle Singularity: Sam Altman’s Blog
- Sam thinks that GPTs are powerful than any human lived on earth, is that really accurate? Maybe, in terms of knowledge, but that’s not truly knowledge. It does have billions of weights that somewhat represent the knowledge, but can it make sense of it? No, not yet. With tools and reasoning, maybe, but not quite without proper instructions.
- The other stuff is just sci-fi future prediction. I don’t think that is true, but his vision is a bit daunting if that is slightly true.
- He also mentions how many watts are consumed per GPT call, which is hilariously alarming.
Anyone can cook: How 37 Signals hired a junior developer
- Intrinsic motivation is greater than an educational degree; this is true for any professional. Because the degree won’t sustain you longer, if there is a fire within, that will carry on.
AI IDEs Free Tier War
- Windsurf is surprisingly cheap, but they don’t have access to models like Claude 4 and others.
- GPT 4.1 is good and all, but that’s a little supbar with the standard of Claude
- I don’t know if that’s just me or I feel confident in copy-pasting a file from claude than from GPT

Watched

Janvi Kalra From Software Engineer to AI Engineer at OpenAI
The Pragmatic Engineer
- This was a really inspirational interview. Its giving me hope in continuing what I am doing currently; being excited and willing to develop stuff, doing the due-deligence to research and solve things.
- I think I now understand my flaw; I am not specific. I am all over the place.
- She was very specific in boiling down what she was interested in, listing down the 50+ companies in that space, and getting the interviews. It shows the care and love for the craft.
- I am not sure how to do that, really, because I think I would miss out on certain roles where I might feel excited. I am really a bad problem solver, I don’t know what I should focus on. I think I need to write more in order to nail it down in the coming weeks. Let’s do that and let’s see where it takes me.

My AI Skeptic Friends are Nuts: Review by Theo
- I read the article last week, and surely it was AI bullish, but the GraphQL, Web3 bubble value to hype ratio just made me relate to this AI hype.
- The hype is too high, but the value bar is high too; we need to get the value, which is a bit easier, but avoiding the hype and fluff is a bit tricky. I think by being hands-on you can separate the fluff from value.

Are we feeling it now? The AI Model Fatigue
- There are too many models to keep track of, I had raised this concern in the past two weeks, but no one listens. Here we are in an LLM model apocalypse.
- Maybe it’s a positive one, but we need to slow down a little.

Dopamine Driven Development
- Passing tests, GitHub action tick, first try especially, different error message all gives dopamine, and that is not a cheap dopamine, its a value to effort cost.

Learnt

Using LLMLite to list different models and providers
- We can use `models_by_provider`, which is a dictionary of provider keys and a list of strings representing model names provided by the provider
Over 3 years of only Python, I didn’t know ABC are abstract classes
- I am ashamed that ABC and abstract_method are a thing in Python
- Never really used them, never made sense, but yes, now it makes sense to use and see them

from abc import ABC, abstract_method

class Shape(ABC):
    @abstract_method
    def area():
        pass
   
    @abstract_method
    def perimeter():
        pass

Tech News

Mistral releases Magistral, a thinking/reasoning model
- It thinks like crazy, I saw from a clip.
Open AI releases o3-pro and drops the prices by 80%
- mini - standard - pro is the ecosystem Open AI has created for its thinking models
- o1 - o3 - o4
- By the time o3 standard or pro was released, o4-mini was already to the same level
Did Apple just give up on AI?
- AI and Apple can’t be on the same boat at least in 2025, it seems.
- Sad to see it, but they are truly behind in tech right now. Their hardware is good, but they have to grow significantly to stay even competitive.

Phew, what a week, nothing much in releases, but the cloudflare x google cloud outage just wrecked havoc in almost half IT companies. Many unicorns faced this downtime for a while. That’s a wild week till June 14th

For more news, follow the Hackernewsletter (#750 edition), and for software development/coding articles, join daily.dev.

Leave a comment

That’s it for this week. Hoped for a good week, got one, hoping for a better week.

Thanks for reading :)

Techstructive Weekly #44

Sat, 31 May 2025 00:00:00 UTC

Week #44

It was week of comeback. 1 videos and 2 live streams, back on track. Though lost the steam in the middle of the week. Still fresh and excited to get hands on projects this weekend.

Appwrite Sites? Android Apps for LLMs? AI Agents with some specific goals?

We’ll roll the die and let’s nature take where it wants to. We’ll start though, I realized it at the end of last week. No brainer right, once I start something, it quickly makes me addicted to completing it and the thing in between is curiosity and consistency. Not hamering buzzwords, but truly, I just started to record the video for pydantic AI and I created a few examples, within an hour I had 7 examples created. Some of which I had to discard, but nonetheless, enough material for a video to record.

Hit the record button and started speaking, some 1-2 hours later, was completed with 1 and 20 minutes of raw footage, next day edited it and published. I t turned out to be a 45 minute video, not bad, it only has got around 30 views in 4 days, but nevermind that. I learned a lot and got a good understanding of AI Agents with Pydantic AI.

A good week, looking to the next week for more projects created and feeling more satisfied with my efforts. Because that is what we can do, focus on the efforts, forget about the results.

Quote of the week

"You can’t connect the dots looking forward; you can only connect them looking backward."

I created the examples and then recorded the video (i.e. the thing I wanted to show). I knew what I wanted to explain in that video, a general direction, but after creating the examples, I really got the understanding of what I actually meant and what I am creating, that was working backwards and figuring it out on the go.

Simialrly in life, we don’t know the future, we can’t. We know the direction, but not the actual destination. We know what we have done, and by doing what you feel right right now is the way to move ahead, overthinking, perfectioning the thing, will just delay it. Just do it, follow the intuition.

Created

Creating AI Agents with Pydantic AI
- Understanding the basics of Pydantic AI
- Using the output type to get the structured response back from LLMs and AI Agents
- Using tools with LLMs called Agents with Pydantic AI
- Using local as well as cloud provider models in Pydantic AI

Live Streamed

Completing the database sync and admin routes for CMS in the Static Site Generator in Golang

On Saturday, I was expecting to create Appwrite sites, but was surprised to find it in a waitlist, so decided to move on and complete the CMS for my SSG

I created the script to sync the database and posts on the github repo, it had a issue with the SQL query and I fixed it later after the stream.

On Sunday, I continued with the CMS part, and added the workflow with the sync script to write the contents to the GitHub repo.

Also added authentication to the Sync DB and Trigger build GitHub actions to only the authors in the site with Netlify Cloud functions.

Finally completing the SSG + CMS in Golang called Burrow.

Read

The Copilot Dillusion:
- This was like a word of caution of getting too much reliant on AI for coding
- Coding with AI is all well and good, but the thinking part still is in the heads of the developer. It can’t think beyond a certain capability.
- Till now, its fair to say from this post that, AI is most safer in the hands of developers than laymen for coding. It’s not gone that far that managers can vibe code and ship everyday.
- Maybe someday it will, but there will be the need of developers in pushing it and nudging it to get the most of it.
Am I online? A useful way to identify your servers are connected to the internet.
Read out the thing you have written:
- This is like a thought or tip from Martin Fowler
- He suggests to read out (at least lip movement) the thing we have written. I really like it and have not noticed it yet, but yes that is true.
- We get a different perspective and feel for the draft that we have written. Its almost like reviewing your code locally vs on GitHub. You get something while reading out that you don’t just reading in mind.
Explaing Transformers in simple words:
- This was a simple yet effective explanation of the transformer model architecture
- It skipped the middle part, it explained the input and output part well though, enough for someone to get curious and fall in the rabbit hole of exploration.

janvi kalra

Explaining GPT-4’s Secret Sauce: Transformers

Transformers are the secret sauce that makes Chat-GPT, DALL-E, and other GPT-based systems so powerful. No, I'm not talking about Optimus Prime and his Autobot pals 🚗🤖 - I'm talking about the neural network architecture. Transformers are the "T" in GPT-4 (Generative Pretrained Transformers v4), and their development has enabled machines to understand …

3 years ago · 13 likes · 1 comment · Janvi Kalra

Let’s Fing Go:
- This was a reflection and a way for switching and learning Go from Javascript. It is evident that Golang is the almost the perfect tool for backend servers and quick simple applications. Not saying its bad for complex applications, but one must be equipped with the right set of tooling and mindset to head into a large scale application, rather than complain about writing bunch of err != nil, skill issues.

Interesting Things

Radicle: The GitHub alternative
- This week we had a GitHub outage, and on twitter I read about someone complaining, how has someone not figured out a GitHub alternative.
- The tweeter mentioned that it was just a

Watched

Inside an LLM
- This 3 videos in the series were so much valuable. It helped me understand the mathematics and the architecture behind LLMs. Its quite fascinating
- I am now thinking about explaining these concepts to laymen because I want them to understand what they are actually interfacing with are just mathematical numbers and nothing human like robots.

MIT Lecture: LLMs introduction
- It had me till the part “LLMs can do math”, no please no.
- All the parts of the videos were great, it touched upon almost everything about LLM and the capabilities it has, helped understand the difference of the actual model and the interface we are interacting with.

API Gateways
- This was a concise explanation of API Gateways
- API Gateways are simply a way to route your app to the microservices you have, and also some gluecode like middleware and rate limiting which might be repeated across most of the microservices.

Primeagent tried AI Video Services
- Just for entertainment, this was funny though (some jokes went over my head though)
- Veo 3 and Luma something was great at almost realistic and relatable videos
- There were services that were clearly there and others were pieces of shit.
- Nothing in between because average sucks.

Learnt

Comparing values in Python. We can also compare any type with the equality operator.
- I thought I might get an error if the value is None and I am comparing the value with a string, but I was wrong
- In Python, the equality operator can allow any type of expression to be compared with any value.

Tech News

Anthropic finally launches the voice mode in its model
- Finally anthropic is adding voice mode. I love the way that anthropic comes at everything from behind and takes everything at it head, hopefully here will be the same.
- The code part, the thinking part, the editing part of code is so reliable in Claude, however the search part I am not sure of.
Mistral launches AI Agents API
- Every company is launching remote agents
- Jules from Google, Codex from OpenAI
- Now mistral, they have similar things
  - Python code environment
  - GitHub integration
Deepseek R1 with the new update to its model
- This seems to be inching slightly up and a less heavier model than devstral or other models runnable in a single GPU or limited RAM
- Let’s see how community tests and where it ranks over the week

For more news, follow the Hackernewsletter (#748 edition), and for software development/coding articles, join daily.dev.

So that’s a wrap from the #44 edition of the newsletter, which has turning out to be my reflection of the week, which it is. Some questions that we can discuss about

Agents are great, still people hate when it is mentioned, what do you think? I meant the tools and LLMs in the loop
Are you using Remote agents while coding? some quick tasks or prototypes?
Are you using local agents like claude code, amp, wrap? how do you handle costs?
What are your thoughts on AI Video generation? do you feel scared about it

Let’s try to wrap our heads around these questions, since we need sometime and mental space to explore the huge impact AI is having on us. TIll then, keep coding.

That’s it from this 44th edition of my weekly learning. I hope you enjoyed it, and leave comments on what you think about some of my takes or any feedback.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #43

Sat, 24 May 2025 00:00:00 UTC

Week #43

What a whirlwind of a week. This was expected, nothing new. New week, new models, every day feels closer to losing the developer job. But everytime, AI sucks and human grug brain works.

This week, I took a step back and tried to ease myself out from this AI slop. Want to really plan the things out and start the grind again. Need to re-align my core values and get back to the consistent and curious flaming torch in guiding myself in the AI abyss.

LLMs and AI Labs are out of control, they are releasing new models left and right.

Number of LLM models released this week: 5

Gemini 2.5 Flash Pro (improved)
Gemma 3n
Devstral
Vercel’s v0-1.0-md model
Claude 4 family (Sonnet and Opus)

This is a new section. At this rate, I am overwhelmed trying to keep track of which model is released at what time, since new models keep popping up constantly.

I spent a lot of time reading and understanding about Agents, and will be creating some form of content around it. For my understanding, as well as creating a resource for this AI hype, to get my hands dirty.

Some personal news, I migrated my blog from markata to my own SSG from scratch, named Burrow. Bold move! I don’t care, no one’s reading my post, so it doesn’t matter, I kept the links intact. A win for me.

Burrow
- Built from bare-bones golang
- Only 3rd party library (goldmark) used for markdown parsing (thought of wiring own markdown parser some day to remove this :)
- Github SDK for some automation and LibSQL driver for database syncing
Markata by Waylon Walker
- Nothing wrong with this SSG, it was fantastic to write our own plugins, and most of the plugins were built-in. But I wanted to explore more about Golang. And this was the move I took.
- The current burrow-ssg in golang is inspired by markata

Quote of the week

"Stay hungry, stay foolish."
— Steve Jobs

I listened to the speech of Steve Jobs, and there were three lessons that he tried to convey:

Follow your heart, let the dot connects
Keep looking, don’t settle
Do it as if you are living the last day of your life

This just are valuable life lessons that he might have spend 50 years on. These are given by wisdom and not knowledge.

Stay hungry, bring the curiosity, and keep fighting and keep throwing your hands and legs at it, you will be at the top one day.

Wrote

Thoughts
- Appwrite is now Fullstack-as-a-Service
- I kind of hate Agentic IDEs for the sake of productivity

Nothing significant, because I have been in the cave, cooking stuff. This weekend it will be something. Pushing hard, coding, not vibe, just bug smashing.

Read

No docs, no bugs
- This is silly; the more detail you put, the more detailed your tests have to be
- The more tests there are, the probability of all of them passing would start decreasing, and hence, it could cause subtle bugs
- So I think it is a matter of convincing yourself to write more tests and documentation
Nothing makes me want to hire someone less than them showing an open-to-work badge
- This is a good take, if someone has put an open-to-work badge, it might mean that the person is giving indication to his/her openness to work.
- We can’t judge by that, and should not, as the person might be in a weird situation and can’t openly express their situation so the badge gives them a slight help in communicating without actually speaking about it.
Revenge of the junior developer
- Ok, I spent half an hour reading this article and felt a mix of opinions
- On one side, I am happy that I am a junior, I am dabbling with AI agents and tab tab things, but it’s kind of weird to use them.
- I don’t like the review part, it feels like a daunting task
I really don’t like ChatGPT’s new memory feature change
- I haven’t used this, but I can understand the feeling of this
- This looks like a feature without a button, a default feature. Really, are these big companies talking about AGI and care about the values, and slopping users with a bare minimum experience?
Build systems
- In short, you should not use AI slop, pretty obvious
  Behind the Screen
  Build Systems
  Everyone around us, even people far from the world of tech and startups, are talking about AI. From social media to mainstream news, WhatsApp groups to coffee chats, there is always some mention of AI…
  Read more
  8 months ago · 2 likes · Yogini Bende
Glauber Costa on Technical Blogging
- These are golden points here
  - Just write it down
  - Showcasing benchmarks is a bit hard
  - If you are not sure about it, reluctant about publishing it, just do it, as you don’t know what will work and what won’t until you haven’t done it.
    Write that blog!
    Glauber Costa on Technical Blogging
    Welcome to our latest attempt to (not-so) gently nudge you to write more! Following up on writethat.blog and Writing for Developers: Blogs That Get Read, we’re sharing the perspectives of expert tech bloggers: why they write, how they tackle writing challenges, and their lessons learned. This time, let’s hear from Glauber Costa…
    Read more
    8 months ago · 2 likes

Watched

Visual Studio Code + Copilot is not open source
- VS Code is trying to save tools like cline and augment code
- Basically, making developers stick to the VS Code ecosystem as closely as possible
- This will make that happen, with Copilot abilities (not the server), access to make the abilities run within VS Code, so people don’t have to fork it.

Google keeps winning
- Gemini 2.5 Pro and Flash are killing it with the new improvements
- Gemini is getting integrated into almost everything at Google
Appwrite Sites
- This is cool, I talked about it in my thoughts as Appwrite is kind of entering the full-stack as a service ecosystem with this launch
- They came #1 on Product Hunt this week
Datbricks bought Neon
- This is a interesting buy
- Data analysis with the powerhouse database ready for AI
- Making a viable integration with the Neon database.
- Great move by Databricks
- Neon doesn’t have enterprise customers, they are low on funds. But they have a wide adoption and great AI integrations
- Databricks has enterprise customers, but not much to offer on the AI side
- Combine the,m and suddenly they both become a deadly combo

Learnt

Using Jules to get things done on GitHub
- Write the thing you want to do
- It will create a branch > you only have to create and merge the PR
Use Open Router to use Anthropic models since it’s not quite reliable with their own API

Tech News

Do we have a website to track which thing, product, model, or tool was released when? If not, I am going to create one specifically for LLM and AI stuff, because this thing is getting out of hand. We are seeing models drop almost every 2 days. This is crazy stuff. OpenAI, Google, Anthropic, Meta, Mistral, xAI, Alibaba, Cohere, in loops over months and weeks with new models. The never-ending cycle. Let’s see if we can predict what model will be next.

Ok, this week we say 5 LLM models are getting released, and those are not any normal models, those are quite a significant push in the state of the art for what LLMs can do.

Anthropic with Claude 4 family (Sonnet and Opus)
- Quite solid in coding task, I just tried giving it a golang file and it corrected all the errors and weird quirks it had.
Mistral with Devstral
- This is a 14 GB, 23 Billion Parameter model for coding task
- Not sure about performance but it requires 32 GB Ram to run, so I am out of uck on that.
- Waiting for a slight lightweight model. Qwen is a great model in that sense.
Google with the improved Gemini 2.5 Flash and Pro
- This was short lived as Claude 4 dropped with all guns blazing
- But Gemini is now getting into everything at Google
- This is leaning on the magnetic route in every product
Vercel with V0-1.0-md model
- This is a bold move from Vercel, could be used by UI developers extensively

Tools

Google with Jules and Stitch
Google’s Flow with Veo 3

Ok, too much Google

Here are the 100 things that Google released at the IO 2025

Phew! That is a lot of news in a week. Google and Anthropic have nailed it. Google is going all over the place with a commanding position, Anthropic taking the focus to the developer ecosystem by making the model better and better at tool calling and coding tasks.

For more news, follow the Hackernewsletter (it’s a Boeing 747 edition), and for software development/coding articles, join daily.dev.

Which model do you think is the ground-breaking model?
What agentic tools/IDE do you find more useful and help in improving productivity?

That’s it from this 43rd edition of my weekly learning. I hope you enjoyed it, and leave comments on what you think about some of my takes or any feedback.

Techstructive Weekly #41

Sat, 10 May 2025 00:00:00 UTC

Week #41

A slow-moving week overall. I had some spiritual realizations, not major, but enough to clear mental clutter and ease the overthinking through action, without worrying about results

The past weekend was too boring; I was not able to do anything fancy. Didn’t live stream, nor create a video. I think videos are not my thing for the moment, It takes too much energy to edit and polish my voice, and I don’t see enough return on that investment right now.

That gives me some time to crunch some articles and blog posts. Will be focusing on that and see if I can get more value than videos, that’s my theory for my skills, but I need to experiment.

AI Slop

Ok, we cannot spend any week without rambling about AI and LLMs, can we? This week, I read a few articles about vibe coding and the jazz people have created that it is going to replace developers. To that, I say: hell no. Yes, it can write code, well-documented even, but can it solve the whole problem? Nope. Not yet! It’s way far from that, however, if you give it all the pieces, and you think the solution and give it nudges in certain directions, then? Yes, to some extent.

We are on a cliff right now, we as developers need to learn to use LLMs to our advantage, but on the other side of the cliff is the part which is almost LLM writing the code, which looking at the speed of accuracy of code being written by different models, its not far but not quite there yet. It is at the stage where only a developer can steer the wheel; once a layman can steer the wheel, developers are doomed, but till then, till the laymen realize that developers are doomed, we are going to rule, take out their money for sure.

Quote of the week

“It is never the tool that decides. It's the hands-and the heart-of the one who wields it.”
― Kevin Sands

AI right now feels like that tool, if you give it to a layman, normal person, they will call it shit for coding, but if you give it to a good developer, he/she will say that I want that in my tool belt always. Creating software is a craft, an art, and it is suited for those learning and caring about the craft and tools, not those looking for expensive or shiny tools. Try, fail, learn, improve, that is LLM at the moment for coding:

Don’t blindly accept suggestions, take time to reflect, it has already saved you a bunch of minutes by writing, may as well read it and understand and think
If you don’t know a thing that it added, search and start a chat with LLM with search documentation tool
If it fails, retry, but think why is it failing, read the error message, if you are experienced, you can fix it quickly than an LLM, don’t waste time and tokens, do it yourself.

In essence, having a general common sense of using LLM tools is vital in today’s software development ecosystem. Might not be relevant to software, but I am not sure.

Read

What even is Vibe coding
- I completely agree with the entire post, this is the best take on vibe coding. I have experienced this and also mentioned it in my post. AI can handle the boring part, so that the developer can focus on the important, the heart of the craft. Whatever AI creates before this needs to be seen with a grain of salt, i.e., needs testing and review, it’s not a real person writing code after all.
- It’s more of a tool than a replacement. LLMs are not there yet, it’s far from reality and close to boilerplate or already solved problems, not innovation.
- If AI is giving you the entire code, then it is likely that the idea that you have for the software is already kind of there, not entirely, but vaguely there. For real innovations in the software, you still would need a developer.
I read the book The War of Art, Why?
- I asked Chat GPT, based on my conversation and everything you know about me, to give me a book recommendation. And it gave the war of art
- And Oh my god! I loved the book, it was a quick one. I don’t usually read non-fiction, self-help books because I already have too many technical things in my work and side projects that I can’t bear reading those in my spare time.
- But that book was a huge shift in mindset. We are all battling resistance. The bigger the resistance, the bigger the calling (the good task or intuition). We have to defeat that enemy, and it’s not easy, but with a war, it can be defeated. And it’s not an enemy once you defeat it, it’s done; rather, it’s a daily war, it will come in the places you are comfortable.
- This prompt was a pretty good use of LLMs, to be honest. If you can tune it to your needs, LLMs are a blessing; if you try using it for your replacement, you are destined to suffer.

Watched

A case against SQL
- I agree to the points that SQL is a bad abstraction of programs, we can use strings to query the database
- We should use the actual interface to access the database to query or mutate.
- But on the scale of complexity and ease of use, SQL outweighs the former
- We can think of SQL as Python and the native way as C; both are viable, but one is easier to write, and the latter is like shooting in the foot.

Learnt

Removing duplicates from a list of dictionaries

def remove_duplicate_dicts(dict_list):
    seen = set()
    unique_dicts = []
    for d in dict_list:
        dict_tuple = tuple(sorted(d.items()))
        if dict_tuple not in seen:
            seen.add(dict_tuple)
            unique_dicts.append(d)
    return unique_dicts

I didn’t code much this week. A bit of slow week, I don’t quite like this type of weeks, but have to go through them. However, I played with Lovable and Bolt for my website overhaul, and Lovable seems to be great, but very limited credits. Bolt is quite clunky and gets interrupted in the middle. very bad experience with Bolt.

Tech News

Gemini 2.5 better model for coding tasks: Gemini is really taking the lead here. Anthropic might lose the bet on the code quality vs cost race here if they don’t launch a great model in a couple of months (We all know they will)

OpenAI’s Sora becomes available for free users for image generation: I heard this yesterday, and I thought it was not there already seems it was there since 1st April. Not sure.
Open AI buys Windsurf for $3 billion: I watched an video of Thoe comparing the usage of these AI IDEs/editors, and Windsurf seems to be low at 4% adoption compared to almost 40% of VS Code and 30% of Other IDEs, and around 20-30% of Cursor usage. Which makes people think, why windsurf? Maybe that is a production-ready enterprise-grade IDE? I have used it and had no problems using it, but Cursor is good too.

Techstructive Weekly #40

Sat, 03 May 2025 00:00:00 UTC

Week #40

What a week, I knew the silence was just temporary, it would come at once, thankfully it was not too much releases to keep track of. There was however one release of a model (or a family of models) from Alibaba Cloud with the Qwen suite of models, that overthrew the Meta AI LLama Models from the SOTA standard for open source models.

This week, I explored a bit about LLMs and how to operate with them effectively, thanks to the prompt engineering masterclass from Anthropic. I created a bunch of things, leaving aside my fear and procrastination. This week, I was a bit specitical on myself but was able to break through the chains and took some hard decisions and executed, the results? Not sure, maybe good, might not work. But that is not in my hands, whatever was, is done. Time to move ahead and explore this AI revolution first hand.

Quote of the week

"Instead of a celebration of everything you know, an anti-library is an ode to everything you want to explore."
— Anne-Laure Le Cunff

I want to do this and this, nope, that might be output oriented

Let’s try to explore this and this, what might we need to do, what we need to go through to feel this?
That is the question that is answered in a process-oriented way.

I want to change this, and I have been reading the book, Tiny Experiments, so far it has changed quite a few things about my procrastination behaviour and helped me take actions over being scared or fretting over the perfectionism bug.

Created

This week, I started a different approach in creating, just publishing at the end of the day.

Built

Published the meta-ai-api tool-call python package
Agentic Calculator: LLM with math tools using Pydnatic AI and Vercel
- LLM: Meta AI Llama 4 (on API Wrapper with tool calls)
- Agent Framework: Pydantic AI
- Deployment: Vercel Lambda Functions

Wrote

Outlined a couple of articles (will publish and reveal soon)
Thought: ChatGPT on WhatsApp → might redirect in some time to → meetgor.com

Recorded

2 YouTube Shorts on ChatGPT and Perplexity released as WhatsApp Chatbots
Livestreamed
- Creating the Meta AI API wrapper with a tool call and a Pydantic AI model wrapper for the Meta AI API
- Building an Agentic Calculator with Pydantic, Appwrite (failed, then moved to Vercel), Meta AI API wrapper with tool calls

Read

Horseless Carriages :
- This was a great read. People making AI-based applications are too naive to understand the actual technology behind LLM and how the approach needs to be changed.
- Right now, it is the assumption that the developer is attached to the system prompt; however, LLMs at this stage need customisation from the user perspective and ndo ot have the dependency on the developer.
- The shift in developer-user responsibility is quite unnoticed, and this article highlights that effectively.
Be Kind :
- As a developer, we need to be kind, not just yes, thank you but really understand the person from other end and be humble about his/her situation and feelings.
- That might be too realistic for a developer to do (right? We are nerds, I think), but having that kindness and insight about the person we are interacting with sets us apart as an effective engineer.
Habits I recommend to a software developer
- Reading a book
- Build projects (keep building)
- Write to reflect the learnings
- Learn and build in public
Getting started with Pulumi
- I always wanted to understand Infrastructue and code, but no article helps me understand what it actually is solving, no one has yet shown the problem before the solution. I might find and write it myself one day (day one?).
Why can’t I be technical
- I didn’t completely read this, but this makes an interesting point, that we need to understand the opposite end of the thinking on the thing we are working on, kind of wired but that makes sense now.

Watched

Lessons on AI Agents from Claude plays Pokemon
- Wow, agents are already playing games, not exactly but quite fair I would say. Anthropic is really a lab, like they are researching LLM behaviors through and through; they are technical scientists.
- Claude plays Pokemon Red:
  - Send screenshot of the current state
  - Describe the game mechanics
  - Ask for the action
  - Iterate
- It’s quite a fascinating experiment. Maybe we can try with different types of games with LLMs. They tried a Pokémon-like game, because that isa very user-paced game, not a very rapid pace, or live-like games. Very smooth transitions and turn based game.

AI Prompt Engineering: Deep dive
- This is a masterclass in prompt engineering. Must watch
- Anthropic really cares about the craft and art of LLMs, they really understand what to think when interfacing to am LLM.
- First principle thinking

Advice for juniors on manager-engineer relation and Theo’s experience (must watch)
- TLDR: Don’t try to do the job that you are not asked, if the environment is pulling each other down.
- It’s doesn’t mean you not following your curiosity or doing the things you are excited to do, rather than knowing when to and when not to.

AI Coding is not enough we need Agnetic Coding:
- This is wild. I have been sleeping on Claude Code and Warp
- Those are the ones that are truly agentic editors.
- AI Coding (which people are pissed at) versus, Agentic Coding, that is really the difference. Agentic coding opens a wild number of possibilities. This is the second time, I am overwhelmed in life in programming. One was with Vim and Linux, there was so many things to learn and experiment with. And this time, its LLMs, models, tools, and so many details to learn, so many behaviours to understand. This is fascinating. Just watch this vide, I can’t be thankful to this person enough.

Learnt

GGUF, the file format for storing LLM model weights
- I wanted to evaluate a idea for a project. Running models from a file, and this file format is what I needed. Using this format and a binding with llama.cpp and other libraries, this can be used for inferencing later to actually run the model
Using llama.cpp python bindings to run a model with a gguf file
- We can use the llama.cpp or other library binding, to load the file in memory and the binding library will be using the inference to get the tokens out from the given prompt

Creating Python Lambda functions in Vercel

The snippet is what you need to get up and running with python serverless functions in Vercel

import json
from http.server import BaseHTTPRequestHandler
from urllib.parse import parse_qs, urlparse


class handler(BaseHTTPRequestHandler):
    
    def do_GET(self):
        parsed_path = urlparse(self.path)
        query_params = parse_qs(parsed_path.query)
        
        query = query_params.get('q', [''])[0]
        result = {"message": "hello, world!"}
        response_data = {"result": result}

        self.send_response(200)
        self.send_header('Content-type', 'application/json')
        self.end_headers()
        self.wfile.write(json.dumps(response_data).encode('utf-8'))
        print(response_data)
        return


    def do_POST(self):
        try:
            content_length = int(self.headers.get('Content-Length', 0))

            post_data_bytes = self.rfile.read(content_length)

            try:
                request_body = json.loads(post_data_bytes.decode('utf-8')
            except json.JSONDecodeError:
                self.send_response(400) # Bad Request
                self.send_header('Content-type', 'application/json')
                self.end_headers()
                self.wfile.write(json.dumps({"error": "Invalid JSON"}).encode('utf-8'))
                return

           
            response_message = {"status": "success", "received": request_body}

            self.send_response(200)
            self.send_header('Content-type', 'application/json')
            self.end_headers()
            self.wfile.write(json.dumps(response_message).encode('utf-8'))

        except Exception as e:
            print(f"Error handling POST request: {e}")
          
            if not self.headers_sent:
                 try:
                     self.send_error(500, f"Server error during POST: {e}")
                 except Exception as send_err:
                     print(send_err)
        return

Using the Warp terminal
- Warp is a terminal and it also has agentic flow, as well nice auto-completion, really awesome.

Tech News

Qwen 3 punches Meta on the face, becomes the sota model in open source models
- Alibaba released the Qwen 3 family of models with almost Gemini 2.5-like results on an open weight model
- The release was almost perfect, the integration so nice, the’ve put effort in making sure the model is available to all vendors on day 1, a state-of-the-art model release, set a standard for release as observed by Simon Wilson and a few others.
ChatGPT and Perplexity make the chatbot available to chat on WhatsApp
- This is a bit of a hot take, it can be revolutionary, it is exposing the LLMs to a wider audience, adoption rate might go high. I am not negative or positive at the moment, but it’s kind of revolutionary to see this things. Open AI again leading this revolution.
Redis becomes open source again
- Wow! A comeback from the tiny database that holds the internet.
- More than that, the v8 release sounds amazing! deep dive next week
Cloudflare launches a Python runtime for serverless functions
- This is interesting, Cloudflare has Python runtime for serverless functions for quite some time now. I found that a few months back. It’s kind of there, and not. As it’s a Pyodide runtime and not a native Python runtime, so there are limitations there. Only a few selected packages are compatible or made available to the pydodide runtime.
- Kind of limiting, since other providers like Appwrite, Vercel, support full-fledged support for Python

Phew — that was a long, wild week. Feels like the whole ecosystem is shifting under our feet. Everything’s speeding up. If I had to bet, next week’s bombshell will come from Google or Anthropic — probably on May the 4th. Just a feeling.

That’s it from this week, will see you in the next one!

Thanks for reading

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #35

Sat, 29 Mar 2025 00:00:00 UTC

Week #35

It has been quite a busy week, with some ups and downs at work. Another Friday release, a lot of patching and end-moment clutches to get it on time. I am not bragging, it is not good, but it is unavoidable at times, I am so proud of that I was able to deliver it, despite having a wired situation in life.

Life is tough, there was a rough day this week in my life, things just get worse and worse, and then get a little hope. That hope might be too tiny, not enough to heal the mind, but enough to carry life. It is tough, I need to stand and stay strong and find a way out because I am destined to be great.

I am receiving signs from the universe about the direction to go, but no matter how hard I try, even from the worse than me, they get results, but for me it just sucks, no response at all. Look at the post I created this week, 0 reactions, man! I put effort and it is not AI-generated, raw human emotion. Tough luck, but someone has said to do the job and not to focus on the result. That is what I am doing and leaving it as is.

Soon, it will change, soon I will be in the bliss of life.

Quote of the week

Robots generate, Humans manifest.

This is a quote I am making myself realize every time I have a problem to solve. I am a software developer, I get bugs, surely LLMs will generate some code, but I want to manifest the code, there is a difference. Humans can show things with emotion and work, AIs can’t do that, humans have feelings, but LLMs won’t have them.

Manifestation is far from generation, it needs a thought, it needs a raw emotion. Writing is not art, it is menial work, crafting stories is art; Generating images is not art, capturing memories, emotion, and expressions is an art; Speaking is not art, connecting, and resonating with emotion, is art; Writing code is not art, solving problems, expressing ideas is. You are a human, whatever you will create will be art. Keep crafting, keep exploring.

Created

I wrote 1 article, published 1 video (finally after almost 2 months), and 2 live streams.

Wrote

A letter for someone who is afraid of AI (including myself)
- I focused on Manifestation vs Generation
- Robots will generate, and Humans will manifest
- Humans need to keep getting good at the craft, leverage AI to make it faster and more efficient

Recorded

SmolAgents: Simple and Lightweight Library for creating AI Agents in Python
I learned a lot, created a few examples
- Basic tools: Multiply and add (call me silly, but LLMs can’t add)
- Specific tools: Scrape a website and list all the links and content
- Answering a user query from a DB
  - Generating the SQL query from Natural language
  - Executing the SQL query as a tool
- Using Local LLMs (llama3.2:1b and Gemma3:1b)
- Using Cloud LLMs (Mistral Small)

Streamed

Porting my Blog to burrow SSG + CMS
Made some tweaks and refinements in the UI and finally it is looking good.

Improving the burrow CMS
Was brainstorming on how to make the edit and delete functionality in the CMS i.e. the sync between the Git and Database

Read

Your manager should never have to ping you: This post discusses that a junior should be asking questions, updating too frequently, or over doing it, I get that point. The author is trying to make sure the manger is never left with suspicion or doubt about the junior “is he working on something?”, “is he even working?”, or “did he understand the problem right?”, because time is precious, and no one wants to waste time, and resources in rabbit holes. Rabbit holes are great but not always and too much of them can deviate you from the problems in a broken way. So I agree with this advice, however, think before dropping the message, don’t try to spam, avoid jargon and I think the communication needs to be to the point. I sometimes get too nerdy or too off-topic in some discussions, that we have to get on a call, yes that happens. No one is perfect, everyone is figuring out life.
Learn to Code: Kache’s Twitter (X) article: This just blew my mind. If someone is optimistic and pessimistic about AI, this guy is just one the extreme end of optimism. Nothing wrong with it. Every coin has two sides, no one can perfectly balance both. But I was kind of doing that, and that was leading me nowhere, after reading this, I am nudged on to the optimistic side. Like I feel there is too much to be done for me, so little time. It just pumped me up.
The worst programmer I know: Yes, this is true. Some engineers or programmers do not actually work on something, but they make something work together. They are the hand-holders, the captains, the leads (not the tech lead, the leaders). That is one type of personality, and that is no right or wrong personality. Humans just are what they are, that’s what makes them humans. This brings me to another point, pair programming, oh! That is a charm, a secret little thing that somehow works wonders. Pair programming is a must, no matter what the scale of the organization is, it opens up the programmer’s mind in different ways, changes the perspectives of individuals, gives confidence, helps learn communication, eases the stress a bit, and makes programming more fun. I can go on, but that is an understatement for pair programming.
My 6 months with the GoTH stack: building frontends with Go, HTML, and a little duct tape: I have heavily done some Golang with HTML for my burrow SSG and CMS. I think at some point, I will have to leverage Templ and Tailwind, I am not a frontend guy, so you know. Templating is really a thing, that when you code in python doesn’t feel but in Go, it actually helps you understand what the heck is going on.
The Friday Deploy
My 6 months with the GoTH stack: building front-ends with Go, HTML and a little duct tape
I’ve been a fan of Go for the better part of 10 years now. Its simplicity and portability quickly earned it a place as my language of choice. But the frontend experience has always left something to be desired for me. While powerful, Go templates …
Read more
10 months ago · 5 likes · Tom Elliott
Deep vs Shallow Go Interfaces: Deep interfaces are really easy and flexible to understand which is really a good observation. The shallow interfaces are the ones that are quirky but unavoidable, we need to have some level of basement laid in order to make it work, otherwise ,it will be all deep interfaces with too many implementation trees. That I don’t like.
We should have more friction in our lives: Cassidy Williams Blog: I shared this with my own thoughts, and yes, we need to bring back some friction as well fiction (pun intended).

Watched

The Primeagen on the Lex Fridman Podcast
Man, that person is a legend.
I have only watched 1:30 hours of this, that is a 5-hour video, man!
- Addiction is something that feels like giving you satisfaction, but it never will, that is a great way to put it.
- Work Harder, not smarter! This is the new advice (instead of working smarter, not harder) Why? because, how would you know something is the optimal way to do it, before doing it? Skewed
Why is Google winning the AI race?
- Google has controlled the internet for the past 2 decades, and it has Data, some model science, and also now has some infrastructure (hardware). It will control the AI race, Apple is actually surprising in all of them but dominates none of them.
- Can apple make a comeback? Looks hard, but we have seen comebacks (Google was struggling with Bard, now it is toppling Open AI)
Git Scraping my own Newspaper
- That’s pretty interesting, git scrapper based newspaper, wow, I want to make this, I want to aggregate my favorite blogs and platforms to get the most up-to-date content. Content these days is overloaded.

Will YouTube ever run out of YouTube Ids?
- Wow, Youtube well thought out, 64 letters, that is a gigantic number, practically it will never run out of ids. However what if someone spams a few billion video uploads?? That can pretty quickly bring the limit close, I mean YouTube will surely block it, but maybe who knows

Learnt

Python’s wired pythonic expression: This I saw from Tsoding’s tweet
- The expression was
  70 * 2, 3
- Expected is `140, 3`
- But why sometimes, my brain thinks it should be 140, 210
- Maybe that is too pythonic, but that is not a tuple expression, It is just a singular expression, where there are two values evaluated
- However, if you do something like this, this makes it more sensible
```
> 70 * (2,3)
(2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3, 2, 3)
 
```
- Creating the tuple `(2,3)` 70 times
- Wired! Sometimes, you don’t have to overthink, just make the thinking a little grounded, there’s too many nuances to keep in the cognitive stack I think.
- Also, this makes it more clear, I think
```
> 70, 2 * 3, 5
70, 6, 5
```
Using smolagents with CodeAgent and ToolCalling agent
I was able to figure out what is the CodeAgent and ToolCalling Agent
- I was trying with the local LLM models those were pretty slow and kind of small (1 billion parameters) compared to the actual 100 billion and god knows how many billion is GPT
- Then I tried mistral on the cloud, generated an API key and everything started to work.
  - The code agent is for generating code, and not explicitly for calling tools
  - There is a separate tool calling agent for that named ToolCalling Agent
  - I tried both of these for their particular use cases
  - Tool calling agent for simple predefined things like scraping the website and getting the links, getting the content, multiplying and add (yes LLMs can’t do that for large numbers)
  - Code Agent for generating SQL queries based on the natural language for the query.
  - That was a great experience, this is what I love weekends for, trying, failing, failing hard, trying something new, trying hard, success. Finally a new error message.
Using default dict with some wired default expression
```
profile = defaultdict(lambda: {"name": "", "links": []})
```
- I didn’t know we could use a lambda expression to create this default value for any key in a defaultdict, I thought that was kind of limited with the built-in datatypes like dict, list, int, boolean, etc. But this opens a wider door for nested data structures. Safe access!

Tech News

A lot of AI News, AI is no more hype, it is a revolution, a moment of life for most of us.

AI News:

Can we breathe for a moment, like there is hardly any time left to check out the releases? At this point, the only way is to adapt and keep struggling and learning new things. Keep the curiosity firing.

For more interesting articles, check out the hackernewsletter for the week edition #339, for even more software development/coding articles, join daily.dev.

That’s it from the last week of March, the end of quarter 1 of 2025. It has been a quick-flowing time, AI is taking over quicker than anticipated, and the pace is quite rapid. In the blink of an eye, a new model, a new improvement, a new tool, is getting launched. Times are wild, and so will the effort.

Keep the curiosity flickering!

Happy Coding :)

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #34

Sat, 22 Mar 2025 00:00:00 UTC

Week #34

It was a pretty busy and productive week. After three hectic weeks of research, hacking, scripting, implementing, Google Sheets, and Python Vulture coding an API, I made a few releases. Yes, that is the normal role of a software engineer at a startup. I get to wear multiple hats: analysing, writing, debugging, scripting, testing, and obviously communicating effectively.

I was able to complete the implementation successfully. The release somehow happened on a Friday evening, which I am skeptical about, but it happened without any hiccups, a few minor ones. It will all be well if it ends well.

Last weekend, I finally launched my own SSG and Blog, which I built with it. I am trying to move the site from meetgor.com to dev.meetgor.com. The site is running on GitHub pages and is built with the Burrow SSG.

A week, productive, and happy about whatever I learned so far. A lot of Python scripting, with pair programming with LLMs, google Sheets for finding accuracy numbers, API shenanigans, and some serious mode all the time. I like being serious, that’s my trait, I am not bragging, just saying.

Next week, I will be hopefully writing more and uploading a video or two about some stuff that I have been tinkering with.

Quote of the week

"It’s not whether you get knocked down, it’s whether you get up."
– Vince Lombardi

I have fallen several times, in the past few weeks, but I have always got up and kept going. It was hard, but facing guilt is more hard than the pain of falling. I can sit and cry, but that won’t get me anywhere, I am here to change my destiny, not to wait and let it unfold. I cannot change winds, but I surely can control the way I move.

Things are a bit different in tech right now, LLMs are everywhere, and they probably can do at par with a junior developer, but one thing is for sure they can’t. The curiosity, the internal fire that keeps a person alive, awake at night, forgetting the surroundings when debugging, this LLM can’t do yet! If you fall, we all do, get up and forget everything that happened, remember you can drive it, drive it better, change the way you react to failure, and don’t let it take over you.

Wrote

Wrote a few things, want to write more.

I will be writing more since I have set up my blog at dev.meetgor.com, it is served by my own Static Site generator using Golang and Netlify Cloud Functions hosted on GitHub Pages. I still need a few things to sort out which will do on the live stream.

I created this SSG in Golang from scratch (except for the markdown parsing) on live streams. This also has a CMS or an API that can add posts to the database and GitHub pages can sync up with the database and save the post to the GitHub repo.

Creating SSG from Scratch in Golang Playlist
Burrow: SSG + CMS in Golang (not a mature and ready-to-use project yet)

Feels really good about how the SSG came together, its really pluggable SSG with flexible JSON as well as YAML compatible frontmatter and configs. It has an editor that creates a post and stores it in the database, the GitHub workflow cronjob picks up the last edited or created post in the last hour and creates or updates the file in the GitHub repo.

Read

AI LLM Agents are just graphs, frameworks over-complicate things
This article shines like gold, it’s quite comprehensive and explains exactly what agents are. The code is precisely 100 lines of code which is surprising to me, was that intentional, or was that a good given luck.
- LLM decides for you
- The tools are rule-based or just procedural steps to execute
- Agents are like graphs, node,s and edges, one pointing to different directions after taking action or deciding what to do next.
  Pocket Flow
  LLM Agents are simply Graph — Tutorial For Dummies
  Ever wondered how AI agents actually work behind the scenes? This guide breaks down how agent systems are built as simple graphs - explained in the most beginner-friendly way possible…
  Read more
  10 months ago · 59 likes · 1 comment · Zachary Huang
Learn Go Templates: A practical guide to layouts, data rendering in Golang
It was a great article. It covered a lot of things in detail as well as digestible examples. I liked the explanation of template partials and functions. I might be using them in my ssg in Golang (burrow). There was one missing nuance, that is using a variable if inside a loop range and referencing with the $ to access the outer variables in templates.

Watched

1. AI is here, and yes you are screwed, as a junior: Article review by THE PRIMEAGEN:

I am here for the hot takes, I read this article last week and shared some thoughts I was completely agreed to the article as I am not sure about certain things, I lack some common sense I think, but yes THE PRIMEAGEN in GOD mode opened my eyes

If you are with people who are working for money, is usually good according to their experiences, which is fair, it sometimes it depends on the interaction with the person rather than the nature of the person. We sometimes get used to the nature of the person after some time.
Use LLM but understand what are you doing exactly
Copilot was great bump in the productivity of developers when it launched but since now we are used to the generated code, it seems like we get bad completion or it’s bad.
A good analogy for LLM's ability to code, the more precise you want to do certain thing, the better the generation is, and the broader the scope of the task, it messes up. Like give me the next 2 lines of code in the current position, that is good, LLM is really good at that (not always). But if you give i want to build this project, in an x and y way with a and b technology, that is where it might get a little screwed (Devin not mentioned but I heard the screams)
AI SaaS clones are shit and not true things, it is more than code, relation, care the bond. If someone is earning money for the SaaS clone they build with AI, then they might be building a silly problem for a silly user.
The point of juniors is that in the future they will pay off the investment they made.
That was too much maybe, a reaction of a reaction, no it’s a note for me.

Backend Banter with Mitchell Hashimoto:

I have watched it halfway through and so far some hot takes

Rust ecosystem is a little messed up to interact with for him at least
Zig fixes the pains of C, the build ecosystem (handles the compilation for us), and package management (like go)
Project-based learning for the win!!
He started to build Ghostty as a fun toy project, then found some pain points and tried to fix those with an ecosystem of libraries to build an abstraction layer of apps on top of terminal emulation.

Learnt

Scrollback editor mode in Zellij
This mode is GOD Mode for Vim nerds, it is really powerful that we can explore the logs from the editor. I used it to grep and search quickly through a long list of logs and filter, count, and extract some things out of the sea of logs. Highly recommend checking it out, might be handy in certain situations.
Format a JSON file in Vim with jq
I always had to open up VS Cod* for formatting the file, which is currently opened in neovim, since I have a low-spec laptop, VSC crashes down the machine, and sometimes also go the browser and pastes in a json frontmatter, but that sucks too. So this command using jq really solved my problem. I use jq frequently when I am outside of neovim, in the shell, however, when I am in the editor, I want it to be formatted immediately.
Using the difflib from Python standard library with SequencMatcher for fuzzy matching
I had been using rapidfuzz and other fuzzy matching tools for string matching. However, there is difflib in the Python standard library, which is really handy for quick simpler things. I found this while asking GPT for a particular thing and it gave me this difflib SequenceMatcher function to compare the two strings, and I was unsure of it at first, but I typed it and surprisingly my LSP picked up difflib, so I checked what is difflib and man that was a python standard library.

Tech News

Open AI releases new Audio Text to Speech API playground → openai.fm
This looks good, really good, podcasts, videos are a bit risky at voice now. Clearly, voice is now taking over with AI, text > images > voice > ??? This is going big.
Claude adds search on the web with generated output (only for Pro Users though)
I have no opinion on that since I have not tried it, but it seems like a great feature and must have since all other major LLMs are masters at this, especially Preplexity, Gemini, ChatGPT, etc.

For more interesting articles, check out the hackernewsletter for the week edition #338, for even more software development/coding articles, join daily.dev.

That’s it from this 34th edition of my weekly learning, I hope you enjoyed it, and leave comments on what you think about some of my takes or any feedback.

Leave a comment

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #19

Sat, 07 Dec 2024 00:00:00 UTC

Week #19

It was a pretty good week, with 2 videos published, 1 short created, and 4 live streams done, and 3 days of AOC solved. Phew!

On Saturday last week, I experimented with Netlify cloud functions in Golang and also for Turso’s LibSQL Database, and that went pretty well. Then started a livestream for the SSG in Golang, and completed the feed loading. Then on Sunday it was AOC Day 1, where I live-streamed it, created the video, and also scheduled the Netlify cloud functions video for Monday.

Then came Monday, derailed for a while but back to business on Tuesday somewhat. The week kept going and I kept solving problems for AoC in Golang and it gave me confidence to improve my problem-solving skills.

Pretty good week and I hope to make it even better next week.

Quote of the week

"Do the best you can until you know better. Then when you know better, do better."
— Maya Angelou

I have been creating videos almost consistently for the past 3 months and I have created 9 videos. I didn’t knew anything about video editing, I was slow at first, but I did the best I could. Now I know some tricks to get the editing done faster I do that to produce more videos and do it quickly. Start → Learn → Improve → Repeat.

Created

Create and Deploy Netlify Serverless Functions in Golang
On the weekend, I experimented with how to deploy a Golang serverless function for the LibSQL database, and through that, I got to understand the cloud function deployment on Netlify with Golang. So, created a video for that on Sunday and published it on Monday.
Live streaming Advent of Code 2024 in Golang → day1 | day 2 | day 3
After completing the basic SSG with Golang, on Sunday I live-streamed solving the day 1 puzzle for Advent of Code 2024 in Golang.
Advent of Code Day 1 in Golang video
After solving the problem for day 1 on live stream, created a short video explaining the problem and solution.

Read

AOC in Golang Day 1: Missing ABS() for Integers
This was a really well-researched and interesting article about why there is no Absolute function for integers. It makes sense now, but I still feel it could handle the general use cases so it shouldn’t be an issue. But yes there is a genuine reason for that to not exist in the standard library.
Command line tools I Like: rwblickhan.org
This was a nice post on the aggregated tools that the author uses most of the time and are quite handy. I also use most of the tools everyday and would like to create a dev container for the setup of my preferred editor/tools and shells and plugins and all that good stuff.
How I stopped worrying and learned to love Go Interfaces
This was well explained and to the point of why interfaces are confusing for most of the developers. They are really handy while working with APIs and I agree sometimes it could be annoying to fit the problems using interfaces, I have come to realize it is not that interfaces are bad, it’s just that we are not clear with the solution to the structure of the program that we are constructing.
The tech Utopia fantasy is over
I haven’t completed reading this post, but till what i have read it feels like we are moving away from the natural and individuality and focusing on the profits and metric games.
Avoid overusing the go init function
It is true, that we overuse certain things like this adding logic in the init functions that makes it hard to test and reproduce the behavior that is expected for a library or a package.

Watched

Advent of Neovim
This is a good video explaining why he uses Neovim and how installing a program from a source helps you to appreciate and understand the project more.
There is also a hidden gem for setting up an appname to a different name to isolate the versions or flavors of Neovim.

Configure Neovim’s options
This was another gem by Teej Devries for setting language/file type-specific configurations or options in Neovim.

Aaron Francis on The Software Huddle
He is a all round good guy. This podcast is really inspiring to learn and be consistent at it, he has gone from Accountant to launching his own company as a developer, speaker, educator, and database master.

Learnt

You can swap variables in Golang just like in Python

a := 5
b := 10
a, b = b, a

This is pretty handy, I encountered this while trying to solve the advent of code problems. I thought can we do this in Golang as I have lazy habits from Python?

Slices.Delete in Golang is Wired
I mean just look at this

s := []int{1,2,3}
slices.Delete(s, 1, 2)
fmt.Println(s)
// [1, 3, 3]

s = []int{1,2,3}
s = slices.Delete(s, 1, 2)
fmt.Println(s)
// [1,3]

I mean, If the function is returning the modified slice, why are we mutating the original one? Do one or the other, not both.

The safer route will be like this then:

s = []int{1,2,3}
newS := slices.Delete(slices.Clone(s), 1, 2)

fmt.Println(s)
// [1, 2, 3]

fmt.Println(newS)
// [1,3]

I need to dive deep into why this is the way it is. Looks pretty confusing to mean.
AFTER A WHILE:

This article explains a bit clearly why the original slice is useless after the operation and we need to pass a copy of that in order to avoid it’s mutation.

Tech News

Amazon’s New LLM Models are the cheapest among the available ones
Open AI launches ChatGPT pro for $200 per month
Bluesky keeps growing in numbers

For more news, follow the Hackernewsletter and for software development/coding articles, join daily.dev

Leave a comment

I have been on bluesky as well, and I must say the engagement and the reach is much better compared to Twitter. I guess I’ll be posting more on Bluesky now.

You can connect with me on meetgor.bsky.social on Bluesky.

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #18

Sat, 30 Nov 2024 00:00:00 UTC

Week #18

It was a great week, if I could call it a hard-working week, it would be appropriate. I started the week (the weekends) by streaming, creating a video, and planning a few more videos for the next week. Then starting the week with deep intense programming sessions, a couple of flow states during the week made the mind feel accomplished and excited.

After releasing the code that I had been working on for so long in the past week, this week was about taking a new approach to the initial implementation, and it was smooth, it felt good, I was able to change the code that I had written faster and iterate on it quickly. This gave me the confidence that I can write production-ready and readable code. I am not that bad, there is a lot to improve on but quite a good feeling to live with and continue the work.

Quote of the week

“The only way to do great work is to love what you do”
— Steve Jobs

This week, I did the things that I loved, working with SSGs, and databases, writing code, and teaching what I know. It might not be great, but these are the things that I love, and other is no one forcing me to do this, this comes out of my will.

Created

VIDEO: What’s new in Golang 1.24: Omitzero Tag in Marshalling JSON
This video was about the specific tag which is being added to the encoding/json package for Marshalling from Golang structs to JSON (serialization.)

TIL: Gotcha with Chained Assignment in Python
I learned this point while implementing a feature at work, I was in a deep flow state, but this thing bugged me and broke out of the state rightly so, otherwise would have gone in wired directions without doing anything significant.
Learning python might be simple, but mastering it is a task, there are soo many things to learn about and be aware about. Python is simple but at times could be dangerous if not careful.

Meet's Substack

TIL: Gotcha with Chained Assignment in Python

I was writing some Python code and wanted to initialize a few variables to an empty list. Instead of creating separate lists for each variable, I decided to use chained assignments like this…

a year ago · Meet

STREAM: SSG from Scratch in Golang (2 streams)
I streamed on the weekend and started where I left off by creating static site generator in Golang. This part was about loading the config and posts from the folders and files.
Then on Sunday, I continued the stream for completing the most basic SSG in golang by adding the rendering of the templates.
- Loading Config and Posts

- Loading and Rendering Templates

Read

Database Indexing Explained: from Computer Science Simplified Substack
This was a great post, explaining the necessary context and fundamentals required to understand the indexing. The last part was a bit heavy, but overall a good and in-depth article on learning indexing.
From laid off to hired: Software engineering guide
This is a good motivational post on getting hired in the era of 2024, this is a new kind of world where you are not only hired based on what you say, but also what your impact says, your brand, your online presence, and actual skills.
How I increased my visibility: Kent C. Dodds
To increase visibility, you need to put it out there. To be seen by others, you have to show your skills and struggles in public. This is to a point article showing practical experience in his journey.
Wafris: Rearchitecting Redis to SQLite
I watched a video on the Database School where a SaaS (Firewall for Web applications) provider moved from Redis to Sqlite. This is such a nice technical post, there are so many things to dive into and how SQLite is the database that just works and not just works but also shines.
Time spent programming is often time well spent
Programming is often more fun than the alternative uses of my time.
Programming is really fun, the joy of creating something and the satisfaction of completing or figuring out by yourself hits different and cannot be replaced with anything else.

Watched

Moving from Redis to SQLite with Mike Buckbee: Aron Francis in the Database School
As mentioned in the above post, this video was absolutely fun to watch, so much technical stuff in a digestible way. There is so much to learn about networking, the quirks of Redis, SQLite and its adaptability, IP Addresses, and interesting problems to solve. Totally worth the hour.
We built an orchestrator from scratch: Fly.io
Why Flyio built their orchestrator, kind of. They use VMs which Kubernetes is not ideal for as it is designed for orchestrating containers, Nomad has a quirk of assigning a minimal number of VMs/machines for the users, which is not secure enough for Fly.io, so reasonable enough that it is worth for them to write and Orchestrator from scratch (well not entirely from scratch)
Why would anyone use Functional programming: The Coding Gopher
Functional programming languages are designed to think about the what and not the how. So we just focus on the what things to implement and not care about how. So, there are things like functions, no variables no mutations (get out of here python).
Let’s Prototype a Javascript JIT Compiler: Andreas Kling
To be honest, I didn’t get most of the technical things, however, I get the point, he is using AI wisely, whenever he uses AI, HE IS IN CONTROL and he doesn’t let AI take over. I have not completed this video, I said last week it would be on my watched list, but I got lost in other stuff and lost interest in the livestream.

Learned

Difference between omit empty and omit zero while marshaling JSON bytes in Golang
From the documentation
The "omitempty" option specifies that the field should be omitted from the encoding if the field has an empty value, defined as false, 0, a nil pointer, a nil interface value, and any empty array, slice, map, or string.

The omit empty is wired, it has its definition of what is empty (which I don’t disagree with but it could be confusing at times). With omit zero, we can control what a zero value is for certain specific data types like structs and time values.
Will write a post on it soon.

How to render templates in Golang
While doing the live stream and building the static site generator from scratch, I explored the difference of Parse and ParseFS which are two different ways of rendering templates.

Template

// templates/hello.html

<h1>Hello, {{.Name}}</h1>

ParseFS

package main

import (
	"embed"
	"html/template"
	"log"
	"os"
)

// Embed the templates directory
//go:embed templates/*
var templatesFS embed.FS

func main() {
	
	tmpl, _ := template.ParseFS(templatesFS, "templates/hello.html")
	data := map[string]string{"Name": "Meetgor.com"}
	_ = tmpl.Execute(os.Stdout, data)
}

ParseFiles

package main

import (
	"html/template"
	"log"
	"os"
)

func main() {
	
	tmpl, _ := template.ParseFiles("templates/hello.html")
	data := map[string]string{"Name": "Meetgor.com"}
	_ = tmpl.Execute(os.Stdout, data)

}

ParseFiles reads templates from the file system, while ParseFS reading templates from an embedded file system.
ParseFiles requires physical files on disk, whereas ParseFS uses files embedded in the Go binary.
ParseFiles allows template modifications without recompiling, while ParseFS requiring recompiling to update templates.

Chained assignment references the same value to all the assigned variables in Python
Python might be easier to learn as said earlier, but mastering it really hard, I leart that chaining assignment is not ideal when you want to iniitialise individual variables to different values (same values but separate on their own moving ahead in the code)
```
a = b = c = [1,2,3]
b.append(4)

# a = [1,2,3.4]
# b = [1,2,3,4]
# c = [1,2,3,4]
```

Tech News

Alibaba Releases QwQ: An Open Source Model as Competitor to OpenAI’s O1 Reasoning Models
Every week, some random thing is happening in the world of AI, someone beats the other buzzing model by little margins. Exciting times in 2025.
Spotify cuts developer access to some developer recommendation APIs
This is bad, and looks like the possible issues moving ahead in this AI driven internet.
BlueSky Open API means anyone can scrape your data for AI training
Everything will have side effects, it depends on how the people are actually using it to their own and others’ advantage.

For more news, follow the Hackernewsletter and for daily developer articles, join daily.dev

Conclusion

This week was all about refinement and iteration, and it felt great to see the progress I made. From streaming and content creation to diving into deep coding challenges, I learned a lot about the intricacies of Golang and Python.

It is Advent of Code from next week, so I am planning to stream daily for solving AoC with Golang in the first iteration and OCAML in the second, let’s see how it goes.

If you want to tune in, slide into Twitch or youtube

https://www.twitch.tv/meet_gor

https://www.youtube.com/@meet-technically

That’s it from this week, hope you did well this week, and have a happy week and weekend ahead!

Leave a comment

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

Thanks for reading Techstructive Weekly! This post is public so feel free to share it.

Techstructive Weekly #14

Sat, 02 Nov 2024 00:00:00 UTC

Week #14

This was an excellent week, I created one video, posted one article, did a live stream setup went live for an hour on Twitch and Youtube, and did some exercise as well—writing of course with the 4th consecutive week.

The week was super-productive to start with, like I had a dream Sunday, I went on a walk in the morning, 6:30 AM, right? I have been procrastinating for months, but finally started doing since this Sunday. And I cannot say how much benefit I had from these walks like there is some energy flowing in me after I come from the walk. There is the fire ignited back, the curiosity flying back, the spark is back and with the habits like this newsletter and others in place, It looks like a dream routine, and it gets better and better. Not gonna Jinx it for the next week.

Stats for the week

Fixed 8 bugs
Created one article post (2k words)
Created a video (20 mins)
Live Stream for an hour (planning a project)
Wrote around 10k words of content

Plans for the next week

Create 2 articles
Record 3 videos
Live Stream at least twice

Quote of the week

"Enthusiasm is common. Endurance is rare."
— Angela Duckworth

This is a good read, motivating to be honest. The start always is the least painful, we quite quickly pick a youtube video to watch but rarely finish it, quickly start with a new year resolution, and falter way too quickly (let’s do at least 2 weeks this year, come on), but the reward on enduring it and keeping up with it is quite a treat.

Read

Using Go Embed: This is really handy and I just used it while researching and creating the video for SQLC. This was really a handy trick and feature in Go, to inject some raw strings or content from files, for instance, it helps in injecting the raw SQL queries from files to the golang program and executes them programmatically. This is commonly used with static files and templating as well I guess.
Taming the beast: Django ORM: Django is a beast when it comes to ORM, but I have started getting back from Django, as I don’t get to go a bit low-level, I can’t write raw SQL. But I am grateful and really happy that I started with Django, it helped me land my first internship and help me understand the terms in the tech. This was the major reason I started blogging and haven’t stopped since then. I went off on a tangent here, but Django ORM is great if you want to do something sophisticated with the data structures that would have in the database.

Created

Wrote: Creating and Working with Embedded Replicas of LibSQL Database hosted on Turso in a Golang Project: I already have created a video on this, so just created an outline and wrote the article on my blog. I find this approach good, it helps me solidify the learnings from the knowledge that I gained from making the video and adding more examples and use-cases.
Video: Using SQLC with LibSQL Database in a Golang Project:
Created a video after 3 weeks, but felt good after creating this. This is the fourth video in the Let’s Go with Turso Series.
Blog: Created a blog for the writing that I have been doing for the past month. I don’t want to share it here, since that is not related to tech. I created a hugo site and added a blowfish theme to it, really got it right on the first try deploy to Cloudflare pages. In to the groove.

Watched

Github Copilot just dropped the bomb on AI Code editor startups:

Learned

Using SQLC to generate go code from raw typed SQL. This is a great tool by the way, I really wanted to write SQL, but the frameworks and tools just make sure you don’t write the SQL yourselves, they generate them for you, which I appreciate that but also they should give the freedom to write raw SQL. SQLC is a really productive and awesome tool for SQL lovers (ORM Haters) and not for ORM lovers (Just kidding, it’s just a tool).
You can embed some content from a file or any place with these `go:embed` tags, this is soo cool. I actually did it for embedding the RAW SQL queries for executing them in the SQLC library.

Tech News

GitHub Universe 2024: There is baggage to be unloaded here. Definitely a lot of AI, but also a lot of other stuff as well. I guess the AI hype is no more hype, it has become a reality, we need to accept it more I guess. But not blindly.
OpenAI launching ChatGPT Search: I mean Bing and Llama almost do this, so nothing fancy here, but the scale with ChatGPT would do it needs to be tested, seems a bit intriguing to be honest. This could be the tool that developers will use instead of Google search maybe in the next year. who knows?

For more news, follow the Hackernewsletter and for daily developer articles, join daily.dev

That’s it from this week, hope you did well this week, and have a happy week and weekend ahead!

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

Also Happy Diwali (Shubh Deepavali, may your life be filled with light soon, you will find hope, just believe) and Happy Halloween!

Techstructive Weekly #12

Sat, 19 Oct 2024 00:00:00 UTC

Week #12

This week, was a bit of a realization that I need to improve as a junior developer, I need to do something important and exciting. These past few months have been tiring mentally, but need to up the effort and get some serious business rolling in.

I will be looking to write more articles based on the videos I create in the upcoming months.

This week, I worked a bit harder than usual, to do my two modes of being. One as a developer, and the other as a writer, I have been writing non-fictional, spiritual, and mythological writing on the side for almost 2 weeks now. I feel really positive, I am able to get things done quickly and appreciate life more gratefully.

Stats for the week

5 bugs fixed
1 Report of experimentations created
4 detailed debugging sessions

Quote of the week

“The way to get started is to quit talking and begin doing."
— Walt Disney

This sums up really well how I have been feeling lately, I need to get things done. I did the exact thing for this newsletter to go live, just write the things I am doing, and it has helped me to realize where I am spending my time, where I am learning and doing more.

I want to inculcate this as a daily routine, to log out stuff that I find exciting, this like won’t be every day, but having that excitement to write things every time I find something interesting is really satisfying.

Read

Lessons from the Smartest developers I’ve worked with:
I completely agree with this point. I feel on the writer’s side, I sometimes come up with weird ideas that don’t align with the business improvement and write shabby saphegetti code on the first try.
From Tech Docs To Blog Posts: I have been writing technical/non-technical articles in tech from almost the beginning of my career. It was another way of re-learning and making my knowledge even firm.
Strategize Your Career
✍️ From Tech Docs to Blog Posts: How Writing Can Transform Your Career
You probably don’t care about tech influencers telling you about their writing journey…
Read more
a year ago · 82 likes · 11 comments · Fran Soto

Watched

What is __init__.py: A basic video, but really important for getting things right when working with organized python code.

Making games, not just games, just anything as long as it is exciting and fun to learn. I had a lot of projects, not any project was big but small individual projects that I made to solve some or the other problems that I faced.

Learnt

Adding a theme to the Hugo site(SSG): It involves a git submodule which if we try to tweak the theme, won’t be committed in the git repo that our own website is being used for. To make changes in the theme, we need to make the fork of the theme and then use it as a separate theme and not as a git submodule.
Postgres has numeric and float types: Numeric for precise and large values, float for loose values but are faster.
Postgres has multi-schema in a single database, as opposed to a single schema type in MySQL. SQLite → single file, MySQL → single server, multiple databases, PostgreSQL → single server, multiple databases → multiple schemas. All have multiple tables within a single database but PostgreSQL can have multiple schemas that we can use to create multiple tables per schema.

Shoutout

Mastering Postgres Course by Aaron Francis: I have been lately watching the course videos of PostgreSQL by Aaron Francis. The quality of content is top-notch, and everything taught is driven by proof, this is really sticking and excellent. HIGHLY RECOMMEND IT!

For tech news, follow the Hackernewsletter and for daily developer articles, join daily.dev

That’s it from this week, hope you did well this week, and have a happy week and weekend ahead!

Thank you for reading, let’s catch up in the next week.

Leave a comment

Happy Coding :)

Techstructive Weekly #10

Sat, 05 Oct 2024 00:00:00 UTC

Week #10

Wrapping up the third quarter with a burst of energy! This week felt more mentally demanding, but as the week closed, I realized that even small progress is still progress. Sometimes, numbers don’t tell the whole story—they can shift perceptions or even mislead. I believe in valuing the journey more than the destination, and this week’s coding, learning, and creating was worth more than any metric can show.

This week, I did some programming, not much, but the weekend was super productive. I wrote an article, created a video, and also learned a lot of things about golang, docker, and SQLite.

Stats for the week

Articles written: 1
Videos Created: 1
Concepts learned: 2
Code Lines Written: ~500
Bugs Caught: 4
Articles Read: 3
Videos Watched: 3
Books Pages Read: 37 (fiction+non-fiction)

Plans for the next week

Creating a video about SQLC library with LibSQL database
Crossposting and cleaning up the article on remote LibSQL database connection in Golang
Contribute to one of the open source projects: Steampipe, MindsDB, Turso, Appwrite

Quote of the week

“Not everything that can be counted counts, and not everything that counts can be counted.”
— William Bruce Cameron

This sums up my week really well. I worked hard to get things up and running, making tweaks, and shifting directions. But in the end, it boiled down to numbers. For me, though, the process was more important than the final metrics. The process helped me reshape my mindset, shift my perspective, and adapt to different conditions as the problem evolved. I suppose I learned to embrace change and gain more insights than the numbers suggest.

Read

Interfaces in Golang: ByteSizeGo: Interfaces are really important in making certain applications, but more important is how to use them effectively and whether to use them or not in your specific problems.
NotebookLM has got everyone in a daze about LLMs and podcasts: I just dropped my portfolio link and it just spits out some compliments after compliments about my journey as a developer. It actually said certain things that I myself haven’t thought of someone viewing me, a couple of things were rubbish for sure, but it was pretty much factual.

Wrote

Connecting a remote LibSQL database hosted on Turso in Golang
This is a write-up for my first-ever YouTube video which after 3 weeks has got around 200 views which is fascinating. Hmm, numbers are wired. After creating a video, the process of writing the article seems so smooth and easy as all the hard work is done, it’s just about to flush out the idea into words.

Created

Setup LibSQL Database server, SQLD locally using Docker and Connecting with Golang:
I also went ahead and created my 3rd video in the series “Let’s Go with Turso” covering how to setup a LibSQL database which is called SQLD locally using Docker and then connecting and querying with a Golang program.

Watched

Tips for handling timezones in Postgres:
It is quite weird and frustrating to manage timezones as a developer it seems.

I have watched very less YouTube for the past two weeks, it is the effect of creating content, consuming less, and creating more.

Learnt

Doker: Mounting volumes with the directory in the system and point it to a location in the container for persistence
To create a mounted volume, you need to specify the host directory path and the directory path inside the container.
```
docker run -v /path/on/host:/path/in/container image-name
```
I used it to mount sqlite db from the local file path to the container of sqld to mount the default db used to serve the libsql database like this:
```
docker run -p 8080:8080 -ti \
-v $(pwd)/data.sqld/:/var/lib/sqld/data/ \
ghcr.io/tursodatabase/libsql-server:latest
```
I am learning about memory management with the course on boot.dev: This course has been made by TJ devries, If you don’t know him, you are missing out on some serious Neovim and Ocaml Shenanigans on Twitter and Twitch. I am not gonna miss this masterpiece while it’s fresh, I know the basics of C, and I should be able to learn the more advanced nuances of memory management.

Tech News

NotebookLM is the buzz, people are creating actual podcasts from certain pieces of aggregated content, pretty impressive to be honest
https://notebooklm.google/
Meta AI launches Movie Gen model that generates realistic audio and video. AI trend seems to be never-ending, each week there is one or the other model being released, taking over the other by the barest of margin but looking like a huge growth.

For more news, follow the Hackernewsletter (#718 has not yet been published but will soon as usual), and for daily developer articles, join daily.dev

That’s it from this week, hope you did well this week, and have a happy week and weekend ahead!

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

Techstructive Weekly #9

Sat, 28 Sep 2024 00:00:00 UTC

Week #9

This week was a bit of refreshing as I was able to write some code, bunch of experimentation and feeling a bit back after some slump in the past week. Need to get one video out this weekend to get into the habit. This October I will be trying to build my own writing routine, it won’t be for technical articles, but I won’t be surprised if I find it helpful to write more technical articles.

This week, I had times where I said, I can still solve problems, Look LLMs you are still idiot. The next day, it just helped me just a function with some function signature and proper data type of parameters and return value. I am not saying, LLMs are complete idiots, but they don’t have sense what they are talking about, they sound like they have, but no, sorry. I would love to improve my productivity, by that I mean to speed up my process for searching, copy pasting from elsewhere, scrambling all over the internet to understand something. It just is like a little pipeline, I throw at it some things, If it makes sense to me, I accept, else I trash it, back to old ways.

I will be making up for the last week slump, I said in the quote

“Courage doesn't always roar. Sometimes courage is the quiet voice at the end of the day saying, ‘I will try again tomorrow”
— Mary Anne Radmacher

Here I am proving it, I am excited to get into the weekend, pumped up and ready for a change in habits and routine. Last week I was frustrated, maybe burned out, I practically didn’t do anything last weekend, but I din’t quit, I said will be back next week.

Actually last week I had my convocation, I remembered it now. The Saturday, the whole day went in that, hanging out with friends, maybe for the last time officially. It feels a bit bitter-sweet that I completed 4 years of the degree college.

Quote of the week

""Fall seven times, stand up eight."
— Japanese Proverb

The past few weeks were tough for me, not as any bad thing, but just things are bit off for me. This week, I still showed up and tackled every thing that came my way, successful or not, I am still standing and doing the thing I love, debugging code :)

Keep walking

Read

How important is Clarity and Kindness in Engineering: Clarity and Kindness, this are really important in any field, but for engineering especially as the dynamics of software development requires patience and focus, which are result of clarity and kindness.
The Caring Techie Newsletter
How to Cultivate Clarity and Kindness in Your Engineering Team
Throughout my career in different Tech jobs - from Software Engineer to Tech Lead Manager - I often faced unclear situations…
Read more
a year ago · 21 likes · 2 comments · Irina Stanescu
Different way of working with SQL in Golang
packagemain.tech
Different ways of working with SQL Databases in Go
Different programming languages have their own ways of working with relational databases and SQL. Ruby on Rails has its Active Record, Python has SQLAlchemy, Typescript - Drizzle, etc. Go, being a language with quite diverse standard library which includes well-known…
Read more
a year ago · 9 likes · Alex Pliutau
Generating music with AI and Python : I actually went ahead and tried a few words to check that out. It actually creates some sound, though it takes time and is bit off track and I actually don’t know what would be the right words for that music that comes to my mind, so I would be experimenting with it just for fun.
The New Era of Writing Code: This is so relatable, I was just using some AI to wire up some functions and making it easier at certain task, but worse at the other, I won’t be surprised to see something like a AI Stack in the upcoming months, Like a set of Models, Providers, templates and prompt cheat sheet for developers and other tasks.
The Developing Dev
A New Era of Writing Code
Large language models (LLM) will change how software engineers write code forever…
Read more
a year ago · 144 likes · 19 comments · Ryan Peterman

Watched

Chat with DHH, TJ Devries and The Primeagen:
This is worth the 2 hours. I would say that much, it is fun and at the same time, so much value for a junior as well any software developer (leaving out the kids part XD)
Mult DB Schema in Turso
Learn how to create child databases from a parent database keeping the same schema, using Turso CLI, this will make any changes on schema of parent db get reflected on the child db.

Learnt

Setup a sqld server with docker: I was checking out SQLD, a server daemon for LibSQL, a SQLite fork database. This is really handy for connecting a sqlite-like database over the network. I am using docker for running it locally and it is a breeze to set it up.
Download LLMs locally with ollama: I was quite impressed by the speed of the LLAMA 3.2 1B model. It worked smoothly, on my 8GB laptop. This would be quite handy in case of network issues, vendor outages and other politics.
LLM package in python by Simon Wilson: This is neat and handy for experimenting and playing around with sophisticated techniques to solve certain problems or even debugging certain hands on problems.

Tidbits of Advice/Quotes/Rambling from others

“College Grads take their work as work, but self taught take it as a hobby, an interesting observation”: Twitter Not trying to quote anyone just agreeing on it.

Tech News

Meta releases LLAMA 3.2 models with lightweight Text (1b and 3b paramters) and multi-modal models(11b and 90b parameters). Read Simon Wilson’s post on his thoughts.
Mark Zuckerberg says Meta AI has nearly 500 million users : That is quite the number, really who is using these model? Is it that good? and how are they tracking these as I know these could be used locally as well.

For more news, follow the Hackernewsletter and for daily developer articles, join daily.dev

That’s it from this week, hope you did well this week, and have a happy week and weekend ahead!

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

Techstructive Weekly #7

Sat, 14 Sep 2024 00:00:00 UTC

Week #7

This week was a bit mentally tough but technically productive, pushed some efforts to get the project up and integrated fully and tested it out. Not good with the numbers but overall, that looked promising, and again excited for next week’s challenge.

This week is special because I created and uploaded my first YouTube video! On Sunday, I did everything from recording to creating the thumbnail, voiceover, and editing. It took about six hours for a six-minute video, but it’s a solid start for future videos. I’d love any feedback you have, so please check it out!

Quote of the week

“You don’t have to be great to start, but you have to start to be great.”
— Zig Ziglar

I created my first YouTube video, it might not be the greatest one, but surely one day one video will be. Pondering over 5 years from now, if that video becomes great, I’ll come back to my first video and laugh—where I started, with a humble and overlooked video. But that will be the one that leads me to that.

Read

One-Year Writing Journey: This is really a good read of a person trying to be consistent at writing, I have personally done this for over two years. I found it really helpful in gaining experience and problem-solving abilities.
Month Without a Smartphone: I could feel myself, but in place of his son, I depicted my mother scolding me to put the phone down. I mean there is nothing urgent or important on the phone, but feels like I am going to miss something important at any hour. This habit needs to be taken care of by me.
Writing a book in the age of Open Source: There are so many good resources and advice on starting and investing in the craft of writing.

Incremental forgetting

Writing a book in the age of open source

About a year ago, Maxim Schepelin and I set out to write our first book. A book for aspiring Engineering Managers…

a year ago · 6 likes · Dunya Kirkali

Wrote/Created

Youtube Video: How to connect a remote LibSQL Databsae on Turso with a Golang Application/Program:

Watched

LLMS are like wired, over-confident Interns:
LLMs are powerful but not quite there yet, they are idiots sometimes, and they literally do what was not said to be done (they can’t push to prod yet)

Software Misadventures Podcast

LLMs are like your weird, over-confident intern | Simon Willison (Datasette)

a year ago · 1 like · Ronak Nathani and Guang Yang

Nobody Cares about Technical GitHub Projects:
Project clones are not gonna help your resume, something that you taught and build just to solve a problem you faced or just writing something from scratch, probably is going to land you somewhere. That is the gist, making GitHub projects is not the only deal, making something that you learn from should be.

Safe and Best Way to Open Files in Python:
I don’t thing anyone uses any other approach than with, for dealing with files in python, if he does then, you probably need to think again.

Why you should learn Golang as a first programming language:
Golang is such a programming language that just lands in the middle of complex syntax and speed. There is a garbage collector which makes it developer friendly most of the time, yet not being interpreted like python and javascript, which makes it faster, not just executing wise but development time as well. Totally worth learning as the first programming language.

Learned

Iterating over a dataframe’s rows using iterrows:
I wanted to iterate over all the df rows and also needed to get the index of each of them, so iterrows is so convenient to use.
Golang: Connecting a remote LibSQL database
- Using libsql-client for golang
- Sync local libsql/sqlite database with Embedded replicas.

Next week will have an article about this things as well as a video.

Tech News

OpenAI released o1: A model series for reasoning and solving scientific problems. People suggesting it to be better than they expected. But I think, the LLMs have disappointed us so much that even a little spike causes us to think its a big deal, it is a revolutionary model, but wait for 6 months and read this blog again.

For more news, follow the Hackernewsletter (Issue 716 is not out yet as of my publishing but should be soon) and for daily developer articles, join daily.dev

That’s it from this week, hope you did well this week, and have a happy week and weekend ahead!

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

Techstructive Weekly #4

Sat, 24 Aug 2024 00:00:00 UTC

Techstructive Weekly #4

It was an exhilarating week for me, learned a lot of stuff, wrote a lot of code, pushed a lot of bugs, and tweeted a lot. This week, I wrote a lot of Python scripts at my work for experimentation setup and testing a lot of things, I was learning and prototyping my idea of Audiofy with Appwrite cloud, and read some inspiring articles.

Let’s refresh the developer’s life a bit and think philosophically.

Quote/Thought of the week

"Learning is not attained by chance, it must be sought for with ardor and attended to with diligence." – Abigail Adams

Not every week is the same, if it is then you are not learning, you are not growing. Stepping outside of the comfort zone is the way to get out of that creative slump. There needs to be hard work and curiosity to drive learning, it won’t come by thinking about how worse the situation is, taking action is the key, if you fail, you learn something.

Read

A Letter to a friend who is thinking of starting something new:
The Sublime
letter to a friend who is thinking of starting something new
Hi hi…
Read more
a year ago · 68 likes · 8 comments · Sari Azout
This blog actually had me think over some of the questions I have by somehow never pondering over more deeply. Because to be honest, I am afraid, those answers will leave me in either fear, guilt, or anxiety. It’s better to answer those and accept them because nothing is perfect in this world.
17 Things you should know about Python Lists to not regret later: As a developer who uses Python at work, I find this extremely helpful, I use most of these almost every day, though some of them also are new to me like list unpacking, not used often much. Some of them are intuitive to use in some cases, which makes Python nice to write.
Go is my hammer and everything seems like a nail: I get that if someone loves to write in a language, it becomes instinct to use it everywhere, but this got me a bit of questioning, like really go for everything? Not sure, like surely could be used, but not necessarily should be used. Would you use Golang to code a frontend application? No right? It’s not meant to be, surely you could write HTML and say I coded a full-stack app in Golang, but really? Been there, done that. Doesn’t feel like touching the code again.
Keeping a daily working journal: I love this idea of journalling daily, I have a prototype of an idea that I made a few weeks back that could be fitted into this domain. Basically an organisation-level blogs for individuals.
Why you should make a New programming language: This blog actually hit me and gave me a hint of my old project of Substack Markdown Parser. So have started to read a couple of blogs and books mentioned in this blog to start understanding interpreters.
Crafting Interpreters: Introduction: As said in the above blog, I am reading this book blog to get some overview of the working of interpreters. I have created a parser for markdown to substack json and vice versa, that is some Python shenanigans. But want to make it a proper parser and interpreter-like tool.
Appwrite: Serverless Function 101 Best Practices: I have been playing with Appwrite’s functions the whole week on the side, it is well documented and this blog added one more silver lining to the rich documentation.

Watched

Appwrite Init Videos:
- Function Local Development:
- CLI Revamped
- Functions:

HTMX CalmCode Crash Course: A really good quick course on HTMX, highly recommend starting here to get a good understanding of why this library exists.
Vim Register and Macros Trick:
This is actually helpful and I knew some of the tricks but being able to get a reminder of that makes me click a few things that I need to change in my daily workflow.
Be a full stack developer with Golang and React: Trolled

Learnt

A list of dictionaries in Python to Pandas Dataframe: This looks quite simple but is really handy and have never thought of it deeply.

import pandas as pd

# List of dictionaries representing bank statement transactions
data = [
    {'Date': '2024-08-01', 'Description': 'Salary Deposit', 'Transaction Type': 'Credit', 'Amount': 3000.00, 'Balance': 3500.00},
    {'Date': '2024-08-03', 'Description': 'Grocery Store Purchase', 'Transaction Type': 'Debit', 'Amount': 150.00, 'Balance': 3350.00},
    {'Date': '2024-08-05', 'Description': 'ATM Withdrawal', 'Transaction Type': 'Debit', 'Amount': 200.00, 'Balance': 3150.00}
]

# Convert the list of dictionaries to a Pandas DataFrame
df = pd.DataFrame(data)

# Display the DataFrame
print(df)

"""
   Date          Description Transaction Type  Amount  Balance
0  2024-08-01       Salary Deposit      Credit  3000.0   3500.0
1  2024-08-03    Grocery Store Purchase Debit   150.0   3350.0
2  2024-08-05        ATM Withdrawal     Debit   200.0   3150.0
"""

Convert a dict with a number as a key and any as a value into a list of ordered numbers with elements as the value of the dict.

# Dictionary with string keys representing numbers and consistent list values
data = {
    "3": [30, 31, 32],
    "1": [10],
    "2": [20, 21]
}

# Convert the dictionary into a list of values, ordered by the numerical interpretation of the keys
ordered_values = [value for key, value in sorted(data.items(), key=lambda x: int(x[0]))]

# Display the ordered list of values
print(ordered_values)

# [ [10], [20, 21], [30, 31, 32] ]

Pip Install a git repo as a python package from a specific branch from the requriements.txt file: I was trying to install the python SDK for Appwrite but the latest release was not live yet, so decided to download the version branch from the GitHub repo. However, that failed as in the Appwrtie function environment, or any cloud/serverless function ecosystem there won’t be git installed.
Appwrite Functions in Python: As I have said, I have been experimenting with the Appwrite function ecosystem throughout the week on the side. I was basically trying to use Python functions for some prototyping ideas. I must say the documentation is really good. I learnt how to parse binary data into a response, a structure for a sample Python function, and set an appwrite json locally for functions.
Table extraction from Images using GPT-4 Vision and Python: This is something I have been researching at my work and found it interesting to read and experiment upon. I am amazed at how beautifully LLMs can parse and get back data.

Tech News

Appwrite Init: It was a great week with a ton of launches and great features. i was really looking for Golang support and it is really a game changer for the developer experience of quickly making a backend with great control.
Redis 8 and AI capabilities: Redis since it changed its license, gets the biggest release ever. As any other thing launching in 2024 has the word AI in it.

For more news, follow the Hackernewsletter https://mailchi.mp/hackernewsletter/713

That’s it from this week, I hope you did well this week, and have a happy week and weekend ahead!

Follow me on Twitter/X for more of my thoughts and tech stuff.

Thank you for reading, let’s catch up in the next week.

Happy Coding :)

LibSQL: Query a remote Turso database with cURL

Wed, 15 Nov 2023 00:00:00 UTC

If you are using a local [libsql](https://turso.tech/libsql) database, it is quite easy to query the database, but for a remote or a database on a hosted cloud platform like [turso](https://turso.tech/), we can use other clients or the api itself to query the data. We can use the turso cli to get the authentication token for the database and then query the database. ## Turso CLI Using the [turso-cli](https://docs.turso.tech/reference/turso-cli) to access the turso platform. We will use turso cli to create a libsql database, create authentication tokens, and query the db. ### Create a database (it's optional, you might already have a database) ```bash turso db create ``` You will get a database on the turso cloud platform with some random interesting name like a passphrase. Use the command `turso db list` and copy the URL ```graphql DB_URL=dbname-orgname.turso.io ``` ### Create an authentication token for a particular database ```bash turso db tokens create db_name ``` Copy the JWT token and this will be used as a authentication token when accessing the remote database in the turso cloud. ```bash TOKEN=abcdef.12345.wxyz DB_URL=dbname-orgname.turso.io ``` * Querying the database using curl or other [api clients](https://docs.turso.tech/libsql/client-access) ```bash curl -s -H "Authorization: bearer $TOKEN" \ -d '{"statements": ["SELECT name FROM sqlite_master WHERE type=\"table\";"]}' \ $DB_URL ``` We can use `curl` or any api client tools to send queries to the database hosted on the turso platform. We need to provide the JWT token in the `Authorization` header to connect to that particular database. The request's body is a JSON string with a list of statements to query the database. ```graphql [ { "results": { "columns": ["name"], "rows":[ ["libsql_wasm_func_table"], ["_litestream_seq"], ["_litestream_lock"], ["sqlite_sequence"], ["user"] ] } } ] ``` The result is a list of key-value pairs as `columns` and `rows` for each of the statements in the body. The columns are a list of column names requested in the query, and the rows are a list of rows where each row is a list of field values from the query.