We are an expanding consultancy looking for an experienced Python Engineer to architect and develop a Retrieval-Augmented Generation (RAG) pipeline. We have a massive repository of unstructured data (mostly PDF reports and internal text documents) and we need an internal AI assistant capable of answering employee questions accurately based strictly on this data, with zero hallucinations. The core stack will be Python. You will need to build the API endpoints (FastAPI or Flask), integrate with a commercial LLM (OpenAI API or Anthropic), and manage the chunking, embedding, and retrieval processes using a vector database (like Pinecone, Qdrant, or ChromaDB). We are not looking for someone to just glue a LangChain tutorial together. We need a developer who understands how to optimize retrieval strategies (e.g., hybrid search, re-ranking) to ensure highly relevant answers. You need to include answers to this question: What vector database do you recommend for this use case and why? 2. How would you handle document chunking to preserve context? 3. Please start your bid with the word "RAG-ARCHITECT" so I know you read this description. NO AGENCY!!!!!