AI Engineer Needed for RAG Chatbot with FastAPI, Vector Search, and OpenAI

We are looking for an experienced AI Engineer to help us build an internal AI knowledge assistant for our insurance operations team. The assistant should answer questions from internal documents such as policy PDFs, FAQs, SOPs, onboarding notes and support documentation. We want a practical RAG chatbot that gives accurate answers, shows source references and avoids guessing when the answer is not available in the knowledge base. This is not a basic ChatGPT wrapper. We need someone who understands document processing, retrieval, prompt structure, vector search, and production API development. We prefer to start with a focused first version, then continue improving the assistant after we test it with our team. What You Will Build? A document ingestion flow for PDFs, DOCX files, CSV files, and simple web pages. A RAG backend using Python and FastAPI. Vector search using Pinecone, Chroma, FAISS, pgvector, or another recommended option. LLM integration using OpenAI, Claude, Gemini, or the best fit based on the project needs. A simple chat interface or embeddable UI for internal users. Answers with source references so the team can verify where the answer came from. Fallback handling when the system is not confident or does not find a strong source. Basic logging for questions, answers, latency, and token usage. Deployment support using Docker and AWS or a similar cloud environment. Short documentation so our team can maintain and update the system. Ideal Experience: Python and FastAPI experience. Hands-on RAG chatbot or document Q&A experience. Experience with LangChain, LlamaIndex, or similar AI frameworks. Experience with OpenAI API, Claude API, Gemini API, or similar LLM APIs. Experience with vector databases such as Pinecone, Chroma, FAISS, or pgvector. Experience processing PDFs and internal business documents. Comfort with Docker, AWS, environment variables, and deployment basics. Clear communication and the ability to share weekly progress updates. Nice to Have Insurance, healthcare, finance, or compliance-heavy document experience. Experience reducing hallucinations in AI answers. Experience adding source citations to chatbot responses. Experience with simple React chat interfaces. Experience evaluating chatbot accuracy with test questions. Expected First Phase Review sample documents and confirm the architecture. Set up document ingestion and vector storage. Build the first FastAPI RAG endpoint. Return answers with source references. Share a working demo with a small set of test questions. Estimated 10 to 25 hours per week. Expected first phase is 2 to 3 weeks. Longer-term work is possible if the first version works well. We prefer clear weekly updates with screenshots, short videos, or demo notes. To Apply, please attach your CV and share one similar AI chatbot, RAG, or document Q&A project if available. Please confirm your hourly rate and weekly availability and also mention the AI stack you are most comfortable with. Simple Screening Questions 1. Have you built a RAG chatbot or document Q&A system before? 2. Which vector database do you prefer for this type of project? 3. Are you available 10 to 25 hours per week?