Gaurav P.

I build production LLM systems: RAG pipelines that hold up on real data, multi-agent workflows that coordinate across tools, and the LLMOps layer that keeps them running after launch. What I build: - RAG pipelines & enterprise knowledge assistants (vector search, hybrid search) - Multi-agent orchestration and AI workflows (LangChain, LangGraph — tool use, memory, planning) - LLM integration into existing products (chatbots, copilots, internal tools) - LLMOps: prompt versioning, evaluation, cost/latency monitoring Tech stack: - LLMs: OpenAI, Claude, open-source models - Frameworks: LangChain, LangGraph - Vector DBs: Pinecone, Chroma, Weaviate - Backend: Python, FastAPI - Cloud: AWS/GCP, Docker If your AI feature works in a demo but hasn't survived contact with real data or a real workflow, that's exactly where I come in.