⁠AI Engineer for RAG & AI Agents for health care | Source-Cited Chatbot (LangChain, Claude) 

We're looking for a senior AI engineer to build a production-grade RAG system with AI agents that makes our document archive queryable with accurate, source-cited answers. The work involves building a retrieval-augmented generation pipeline over our PDFs and scanned documents. We need OCR-based ingestion with structure-preserving extraction, hybrid retrieval with reranking over a vector database, and a conversational AI chatbot that answers questions with citations back to the exact source document. On top of retrieval, we need an AI agent layer that validates extracted fields, flags missing data and discrepancies, and routes uncertain cases to human review instead of guessing. Accuracy is non-negotiable, the system must never fabricate numbers, codes, or figures, so every answer has to be grounded in retrieved content with anti-hallucination safeguards throughout. What we need built: •⁠ ⁠Document ingestion pipeline with OCR fallback for scanned files •⁠ ⁠Structure-preserving extraction and chunking with metadata •⁠ ⁠Embeddings into a vector database •⁠ ⁠Hybrid (semantic + keyword) retrieval with a reranking layer •⁠ ⁠A multi-agent workflow (LangChain / LangGraph) for validation, discrepancy detection, and human-in-the-loop review •⁠ ⁠A source-cited chat interface where every answer references its source Ideal candidate has shipped accuracy-critical RAG and document-intelligence systems in production, with strong command of Python, LangChain, AI agent development, vector databases, and OCR pipelines. Clean, documented code and clear communication matter to us. This is a one time project but might turn into a long term contract if the situation allows. Please share relevant RAG / document-intelligence / AI agent work when you apply. Also share your availability hours each week.