AI-Powered Document Q&A Assistant (RAG-based)

We're looking for an experienced developer/AI engineer to build a web application that lets users upload documents (PDF, possibly DOCX) and interact with them through a natural language chat interface, powered by a Retrieval-Augmented Generation (RAG) pipeline. Core Requirements: Frontend: Clean, modern UI (React/Next.js preferred) with a document upload panel and a real-time chat interface Backend: REST API (FastAPI/Node) to handle document ingestion, text extraction, chunking, and embedding generation Vector database integration (e.g., FAISS, Pinecone, or ChromaDB) for semantic search over document chunks LLM integration (OpenAI, Claude, or open-source models via LangChain/LlamaIndex) to generate context-aware answers Support for multi-document upload and management (list, delete, switch between docs) Conversation memory so follow-up questions retain context Error handling for unsupported file types, large files, and failed extractions Nice to Have: Source citation/highlighting — show which part of the document an answer came from Adjustable chunk size / retrieval settings Authentication and per-user document storage Deployment-ready setup (Docker, environment configs) Deliverables: Fully functional web app (frontend + backend) Basic documentation on setup and API endpoints A short demo walkthrough Ideal Candidate: Has prior experience building RAG pipelines or LLM-based chat applications, is comfortable with vector embeddings and prompt engineering, and can explain technical tradeoffs (e.g., chunking strategy, retrieval accuracy vs. speed).