Sabitlenmiş Tweet
Krishna
902 posts

Krishna
@KrishXCodes
21 | Building in AI | Founding AI/ML Engineer - https://t.co/HMT8pfhiq4 | Super30 - @100xdevs
India Katılım Mart 2021
776 Takip Edilen1.4K Takipçiler

My memory engine hits a vector DB each session.
Worked great until memories started contradicting each other.
The problem?
Duplicate memories with slight variations were both getting retrieved and injected into context.
Here's how the dedup actually works:
Vector search finds similar memories above 0.8 similarity. Those get passed to an LLM alongside the new facts.
The LLM acts as judge -
it decides: ADD, UPDATE, DELETE or NONE.
Returns structured JSON.
Then hardcoded logic reads the event field and routes each memory accordingly.
AI makes the decision.
Code executes it.
Clean separation.

English

@KrishXCodes man this project
figma to react
idea is insane !
good luck mate!
English

My Figma-to-React pipeline converts screenshots section by section.
Each section gets its own agent.They all run in parallel. A supervisor validates the output.
But here's the tricky part:
Screenshots have compression artifacts. Colors don't match the design file exactly.
So I had to explicitly tell the model:
"convert JSON colors to hex, ignore visual noise."
Context engineering isn't just prompts. It's knowing exactly where your pipeline breaks.

English

@stephen_fr_mx Appreciate it! That mindset works, keep building XArmy 🚀
English

@KrishXCodes 1300 in 15 days just by sharing what you build is crazy good. Proves you don't need tricks, just real stuff people care about. Building XArmy right now with the same mindset — ship, share, repeat.
English

@KrishXCodes Those who celebrate on your behalf are only a few. you are real one bro.
English

@priyankapudi This is the stuff you only understand after production humbles you 😭
English

You're building an app from scratch?
Before writing a single line of code, understand these:
> How your data model shapes everything downstream
> Why replication strategy matters before you hit scale
> The real cost of choosing eventual consistency
> That caching solves speed but creates a whole new class of bugs
> Encoding formats decide if your deployments break users or not
> Backpressure > crashing under load
Most devs learn these after production goes down.
Learn them before.
English

Reality of IT Jobs Nobody Tells You:
- Degree alone is NOT enough
- Skills matter more than marks
- First job salary is usually LOW 😔
- Freshers feel confused at start
- Not all IT jobs are coding
- You will Google things daily
- Bugs can ruin your whole day
- Meetings waste a lot of time
- Deadlines give real stress
- Long screen time hurts eyes & back
- Overtime is common, not rare
- Night calls for global clients
- Work-life balance depends on company
- Your manager decides your peace
- Office politics exists everywhere
- Learning never stops
- Technology changes very fast
- Mistakes are part of learning
- Pressure builds up slowly
- Mental health matters in long run
- Experience > certificates (mostly)
- Weekends go into upskilling
- Hard work ≠ promotion always
- Job security is never guaranteed
- Layoffs can happen anytime
- Switching jobs is normal in IT
- Salary growth takes patience
- Pressure can kill passion
- Consistency beats talent in long run
- Skills matter more than degree
- First few years are about struggle
- Growth depends on choices you make
- IT needs self-discipline
- Comfort zone kills career growth
- Networking matters in long term
- Learning outside job is necessary
- Comparison kills motivation
- Burnout is more common than you think
- Discipline beats motivation
- Adaptability is the real skill
English

@kapilansh_twt True if you skip understanding. AI should speed you up, not replace thinking.
English

vibe coding is just a fancy term for
"I have no idea what my codebase does"
→ AI writes 400 lines
→ you don't read it
→ it works
→ you ship it
→ 3am production fire
→ you have no idea where to start
→ ask AI to fix it
→ AI breaks 3 other things
we're not building faster
we're just breaking things
at the speed of light
and calling it innovation
English















