Anish

what punishments of god are not gifts ?
Building @hirebydevtrace

[Anish @anishh_3]

Vector search is not enough.

If your RAG system struggles with error codes, exact identifiers, or rare technical terms, BM25 and hybrid retrieval might be the missing piece.

A breakdown of BM25, Hybrid Search, RRF, and Ensemble Retrievers. https://t.co/Sl1Z7e8pXE

[Anish @anishh_3]

I think synthetic data is going to become a much bigger deal than people expect.

Not because real data stops mattering.

But because a lot of teams simply don’t have enough clean, labeled, edge case heavy data to train or test on.

The interesting part won’t be generating more data.

It’ll be generating data that is actually useful.

[Anish @anishh_3]

@yourstrulysaeed Right

[Anish @anishh_3]

Is it worth doing mtech?

[Anish @anishh_3]

@debuglol 🫡

[Anish @anishh_3]

@duke_silver_ag ?

[Anish @anishh_3]

@KASHIFALIKHAN94 Yep

[Anish @anishh_3]

If you’re starting with vibe coding, do this:

-> use it for small projects first
-> make the AI explain unfamiliar code
-> review every file it changes
-> keep your stack simple
-> commit often
-> test after every major change
-> learn enough fundamentals to debug without AI
Vibe coding is insanely useful.
But if you can’t tell when the AI is wrong, you’re not really in control of the codebase.

[Anish @anishh_3]

@pablomanjarress Haha true

[Anish @anishh_3]

@KrishAdak2 Right

[Anish @anishh_3]

@DevChoudhury14

[Anish @anishh_3]

@_musaqureshi 🫡

[Anish @anishh_3]

@storm_191 🫡

[Anish @anishh_3]

@MasonBattery Thanks bud 🫡 yes I'm interested in going more deep in aiml field but not for job or research, I just want to learn stuff

[Anish @anishh_3]

@Pav192099 Yeah kinda interested in aiml

[Anish @anishh_3]

@vedpmaur 🫡aight

[Anish @anishh_3]

@anurag_x34 Oh atb for it !

[Anish @anishh_3]

@vedpmaur Let's remove the college factor , now tell me

[Anish @anishh_3]

@anurag_x34 Idk saw alot of opinions about it , most people think it's not worth doing because of ai coming up , entry level jobs going etc etc

[Anish @anishh_3]

@rag_rro Great take bud

[Anish @anishh_3]

I think we’re overcomplicating AI products a bit.

Most people don’t care if your app has 6 agents, memory, MCP, RAG or whatever.

They care if it actually saves them time and doesn’t randomly break.

The boring reliable AI products will probably make more money than the flashy ones.

[Anish @anishh_3]

@purvaasays I mean why not