satya pramod

Senior Full Stack Engineer (10+ years, Node.js/React/Python) now building production AI systems — LangGraph agents, RAG pipelines, and LLM-powered applications that ship, not demos. What I've built recently: → A LangGraph ReAct agent with autonomous tool orchestration, dynamic Claude Haiku/Sonnet routing by task complexity, and full LangSmith observability → A hybrid-search RAG pipeline (pgvector + BM25 + reciprocal rank fusion + Cohere reranking) achieving 0% hallucination rate through correct model abstention — the system says "I don't know" instead of guessing when retrieval comes up short → An AI application assistant using tool-based structured output validation, evidence-backed reasoning (not keyword matching), and resilient multi-model fallback design What sets this apart from tutorial-following: I debug the parts that don't show up in demos — what happens when an API is overloaded mid-request, how you validate a model's structured output is actually trustworthy, why a "perfect" test score is usually a bug hiding in plain sight. 10+ years of full-stack engineering (Node.js, React, Python, FastAPI) underpins all of this — I'm not new to shipping production systems, just applying that discipline to LLM engineering specifically. Open to Applied AI Engineer / GenAI Engineer / LLM Engineer roles, particularly teams building real products with LLMs rather than research.