Senior AI/RAG Engineer to Finish Production Retrieval Build (Python/Flask)

IlmBase is a live production AI research tool that retrieves answers from ~8,400 classical Arabic source texts and returns cited answers in English. Stack: Python/Flask, vector search, on DigitalOcean. I'm hiring one engineer to finish the V3 build ahead of a September deadline. The hard part is retrieval quality, not building a chatbot: reducing retrieval variance, keeping sources from the four Sunni schools from bleeding into each other, and making the system flag when the classical evidence is too weak or divided to answer confidently rather than producing a smooth wrong answer. Please answer in your reply: 1. Is Python/Flask a primary part of your stack? (You'd own an existing Flask codebase, not rebuild it.) 2. Will you personally do the engineering, or would work go to a team? 3. How have you handled retrieval reliability: hybrid search, reranking, hallucination/grounding evaluation? A concrete example helps. Strong candidates will be invited to a small paid test task (a few hours, fully paid) before any larger engagement.