I had a conversation with a restaurant owner who needed a simple POS system along with custom analytics to better understand the business and manage operations remotely from home or abroad.
Using AI, prompts, and a clear understanding of the problem, I built a working solution.
Built a small RAG pipeline to understand where things actually break.
π https://t.co/QSq1ZQ80Ip
- Chunking > model
- Top-k can hurt
- Embeddings = re-index pain
- More context β better
- Eval is still unsolved
#RAG#LLM#AIEngineering#SystemDesign
I was reading @FastAPI docs and found a new feature: "Vibe Coding" π
@app.vibe("/vibe/", prompt="return hello world as json")
async def health_check(body: Any):
Tried it β AttributeError π
Yep, April Foolsβ by FastAPI team β and itβs too real π
No guesswork.
No scattered records.
Just clear data β clear decisions.
From the 1989 analog era to the AI era Barnick Pracharani
Building custom application for family business.
This days, noticed that after awaking up I got into reels.
- got the problem
- thought some solutions for me
- will work on that tomorrow
Gonna make a medication and personal healing type of thing.
@ethanmonkhouse Mostly the quality of feedback. I expected pushback, but a lot of people related to the problem and shared real use cases.
Some personally messaged about the solution and showed interest.