This project implements a multi-level Retrieval-Augmented Generation (RAG) chatbot that answers questions based on the content of uploaded PDF documents. It uses semantic search, vector databases, and large language models (LLMs) to provide accurate and contextual responses.
- 📄 Upload and parse PDF documents
- 🔍 Semantic search using embeddings and vector stores
- 🧱 Modular architecture using LangChain
- 💬 Conversational memory for multi-turn interactions
- 🏷️ Metadata tagging and filtering
- 🤖 Agent-based orchestration with tool routing
- 🌐 Streamlit-based web interface
- Python 3.10+
- LangChain
- FAISS
- Sentence Transformers
- Streamlit
- PyMuPDF
- Hugging Face Transformers