Meet Gor

Techstructive Weekly #54 GPT OSS, GPT 5, Python tidbits, learning about SQL IIF, sequence tables, CTEs among all the other things read, watched and created for the week 3rd to 9th August 2025

Newsletter Week 54

A lot of LLM model releases, learning more about SQL sequence table, python gotchas, sqlc bug fixes, sqlite exercism track solving and more

techstructively.substack.com/p/techstruct...

This is so relatable, I haven't done cfps but have started a newsletter (54th edition today) and yt channel

It helps me get things that I love, want to talk about out of my head, and deepen the knowledge

I do similar newsletter but for things I learnt, watched and read (not all but the great ones)

Don't Tell Engineers What to Do A famous example where telling Engineers what to do backfired was the Space Shuttle Challenger disaster in 1986. Engineers at Morton Thiokol, the contractor responsible for the shuttle’s solid rocket ...

Such a mind soothing article, made me feel relaxed after the last 2 week dilemma at work🤌

well written @dblockdotorg.bsky.social

code.dblock.org/2025/07/30/d...

reader.waylonwalker.com/reader_waylonwalker_com/plugins/reader.py at main · WaylonWalker/reader.waylonwalker.com Contribute to WaylonWalker/reader.waylonwalker.com development by creating an account on GitHub.

Wow, this is so simple, people pay for this?
I saw a few apps on playstore most of them offer few sites for free but then have paid versions for additional feeds

Nice work here

github.com/WaylonWalker...

Great, this looks good

Will be trying out to create my own over the weekend.

I actually thought getting content from posts was not ethical but RSS is a thing, so it makes sense to have a read later app combined with rss feeds

Meet_Gor - Twitch Creating Mozilla's Pocket alternative FOSS version

Live streaming

Go and some SQL

www.twitch.tv/meet_gor

Wrote a few thoughts on kubernetes.

Both are opposite and I agree to both of these, as usual in tech it depends!

www.meetgor.com/thoughts/kub...

www.meetgor.com/thoughts/jus...

Need to start using a rss reader, is there an android app for such, any recommendations?

If not i'll later be making one

Learnt a bit of tmux scrolling with vi like keybindings, this will help me do printf debugging much more effectively then using a debugger😅

meetgor.com/til/tmux-scroll-mode-select-and-copy

Techstructive Weekly #52 One year anniversary post! SQLite internals, DHHs wisdom, AI Assisted IDEs and Coders are taking over, celebrating this feat with a happy client in the end, with summary from 20th to 26th June

Wrote the 52nd edition of my weekly summary of tech things i do.

Completed 1 year in doing so

open.substack.com/pub/techstru...

Techstructive Weekly #50 Learning more SQL, Postgres, trying harder, need to change the lifestyle with new habbits or workplace, among the other things read, created, and learnt from the week of 6th to 12th July 2025

Wrote the 50th Edition of my weekly newsletter (all. things I learn, create, read and watch)

open.substack.com/pub/techstru...

Some motivation to carry on

This is nice!

I read something opposite last week and found this surprisingly positive, I will share the thoughts soon

Which LLM writes the best analytical SQL? Can natural language replace SQL? We benchmarked the SQL-writing ability of the top 19 LLMs to find out.

SQL is still a skill

I am a bit surprised with this, I thought LLMs would give near perfect queries (they are close) but meh!

Thanks @tinybird.co for the detailed report, shows LLMs are not quite there yet, SQL is still worth learning

www.tinybird.co/blog-posts/w...

I lost the tweet or something, but I read somewhere

"AI is thinking fast, so we need to think slow"

What a quote, who quoted this?

Listing the supported ai llm providers and its models from litllm

There are almost 900 llm models on litellm package to use from

I have barely lived that many weeks in my life

I had written about it, couple of months back, on the similar note:
t.co/yc3u6k1j0I

Great to see tools like procreate support human created art than diving into AI hype

In the midst of a revolution, it feels humans are drifting away from intrinsic art to generative art, which doesn't give gratification

t.co/N1Yv3a775R

Practical advice

But getting over that dual feeling of they wants to see this vs I want to build this, a tussle which spirals into me creating nothing

This is therapy, I needed this😌

I was just watching and reading about under the hood of LLMs working and it made me feel good (cause its all math and probability)

But the way in which people and companies are taking stances is quite scary.

YouTube video by MeeTechnically Creating AI Agents with Pydantic AI

Created a video on agents with @pydantic.dev ai
- Creating bare bones agents
- Adding tools
- Using different llm models (local and cloud)
- Running agent in different ways (a
sync/async, streamed, graph)

youtu.be/_m2YpvsdxSA

18th to 23rd May 2025

- Gemini 2.5 Pro and Flash (improved)
- Gemma 3n
- Jules and Stitch (tools)
- Anthropic Claude 4 family (Sonnet and Opus)
- Devstral
- Vercel's V0-1.0-md model

At this rate, we have new models every day, hard to keep track of, exciting yet overwhelming

If we create a word cloud based on AI tools and products/models we would quickly run out of space at this point.

AI bros are cooking hard

3 models per week
5 tools per week

At this rate, we'll need to develop a mind map of thinking which tool to use when

Meet Gor Tech blog by Meet Gor

Fresh look on my blog

Now built with Go 🚀

Custom ssg from scratch
- admin mode
- turso database

www.meetgor.com

This is hilarious, I thought it was written by human at first, then got it
(I am still confused, is it written by AI, or are those two author AI Agents?)

quarter--mile.com/Human

What a weather in Mumbai boy!

Dark clouds, complete black out
Feels like 6PM on 11AM

Is it a me problem or does anyone experience?

While reviewing LLM generated code
Like seniors might catch errors or mistakes quickly

But for juniors and smol brain devs like me haven't developed an instinct to review code (human), the moment they were developing, LLMs took over and are we screwed?

Created a silly side project, to explore Agents and tool calling

- Meta AI(api with llama 4) for the LLM
- Pydantic AI as the agent framework

(Purpose? To avoid GPT hallucinating in calculations)
I has math tools (not all) to use before answering

agentic-calculator.vercel.app

GitHub - mozilla-ai/any-agent: A single interface to build and evaluate different agent frameworks A single interface to build and evaluate different agent frameworks - mozilla-ai/any-agent

At this point, we have abstractions of abstractions

Agent framework to avoid the abstraction of all the other ai agent frameworks

Where are we headed?
github.com/mozilla-ai/a...

Wow! This blog made me understand what developer advocacy really is.

I always thought it was making the developer experience nice, but it's more than that.

Really well put in words.