AI-Powered Document Question-Answering System Using RAG

Project Requirements Users can upload PDF, TXT, and DOCX documents. The system automatically extracts text from uploaded files. Extracted text is divided into manageable chunks. Embeddings are generated and stored in ChromaDB or FAISS. Users can ask questions about the uploaded documents. The system retrieves the most relevant document sections. An LLM generates answers using only the retrieved context. Each answer displays its source document and relevant page or section. If the information is unavailable, the system responds with “Information not found in the uploaded documents.” A simple and user-friendly Streamlit interface is required. The application supports multiple document uploads. Users can clear or reset the conversation. API credentials are securely managed through environment variables. Basic validation and error handling must be included. Suggested Technology Stack Python Streamlit LangChain or LlamaIndex OpenAI API or a Hugging Face model ChromaDB or FAISS PyPDF and python-docx Deliverables Complete working source code requirements.txt .env.example Setup and installation instructions Sample documents Application screenshots A short demonstration video GitHub-ready README.md Basic testing and error handling