Build an Enterprise AI Knowledge Assistant (Python + FastAPI + RAG + Local LLM)

Build an Enterprise AI Knowledge Assistant (Python + FastAPI + RAG) We are looking for an experienced Python AI developer to build a modern Retrieval-Augmented Generation (RAG) application using local open-source models. The application should allow users to upload documents, search their contents using semantic search, and chat with an AI assistant that provides accurate answers with source citations. This is a portfolio/demo project, but the code should follow clean architecture and production-quality standards. Technical Stack Python 3.12+ FastAPI LangChain or LlamaIndex ChromaDB / FAISS Sentence Transformers Ollama (Llama 3.2 / Mistral) SQLite Docker React or Streamlit Required Features Backend FastAPI REST API Modular project structure Async endpoints Swagger/OpenAPI Logging Environment variables AI / RAG PDF Upload DOCX Upload TXT Upload Automatic document chunking Embedding generation Vector database indexing Semantic search Context retrieval AI response generation Source citations Chat Features Chat interface Streaming responses Chat history Conversation memory Follow-up questions Source Citation Every answer should display: Document name Page number (if available) Relevant text source Optional CSV support Excel support OCR Image search Admin dashboard Multiple knowledge bases Deliverables Complete source code FastAPI backend Frontend Docker setup README API documentation Sample documents GitHub repository Proposal Requirements Please include: Similar RAG or AI projects. GitHub profile. Which vector database would you use? Which local LLM would you recommend? Estimated completion time. ⭐ Bonus (Will Increase Your Chances) If you have experience with: MCP (Model Context Protocol) AI Agents CrewAI LangGraph Hybrid Search Redis PostgreSQL + pgvector please mention it in your proposal. Screening Questions Add these in Upwork: 1. Have you built a RAG application before? Please share GitHub or screenshots. 2. Which vector database would you use and why? 3. Which local LLM do you recommend (Llama 3.2, Mistral, Gemma, DeepSeek, etc.)? 4. How would you structure this project for production use? 5. Can you complete this within 5–7 days? Attach a Reference Image When you reach the attachments section, include a simple architecture diagram like this in the description: User │ React / Streamlit UI │ FastAPI │ ┌──────────┴──────────┐ │ │ Document Upload Chat Request │ │ └──────► RAG Pipeline ◄──────┐ │ │ Chunk Documents │ │ │ Generate Embeddings │ │ │ ChromaDB Vector Store │ │ │ Retrieve Context │ │ │ Ollama (Llama 3.2/Mistral) │ Answer + Source Citations