Yash Raj

I'm an AI engineer who builds production LLM applications, RAG systems, and agentic AI from scratch. Not tutorials. Not demos. Actual products that are live, deployed, and being used.

In the past year I independently built and shipped four AI products. SupportSnap generates fully functional AI support chatbots from any website URL using a complete RAG pipeline web scraping, semantic embeddings, vector search, and LLM streaming in under 60 seconds. PatientBridge is a medical AI platform that generates structured clinical summaries from patient symptom inputs before consultations begin. DBDesign converts plain English descriptions into interactive ER diagrams and production SQL instantly. GitAuto is a LangGraph agent that reads GitHub issues and opens pull requests autonomously — it indexes your entire codebase, reasons through the fix using RAG, and ships the code without a human in the loop.

My core stack is Python, FastAPI, LangChain, LangGraph, pgvector, and Gemini and Groq APIs. I understand the complete AI engineering pipeline from chunking strategies and embedding models to agent tool design, LLM streaming, and cloud deployment on AWS and Railway. On the professional side I've built real-time API infrastructure for an AI-powered calling platform at NexTalk and worked on a product serving 5M+ monthly users at NoBroker Technologies.

What genuinely interests me is the space between AI that looks impressive in a demo and AI that holds up in production. Bad retrieval, hallucinated outputs, agents that loop forever — these are real engineering problems and I find them more interesting than most people do.

I'm based in India, completely comfortable working async across US and EU timezones, and actively looking for a remote AI engineering role at a startup building something real with LLMs. I move fast, I own my work end to end, and I'll figure out what needs to be built without someone pointing me at a Jira ticket.