Technologies Used: Streamlit, LangChain, FAISS, Hugging Face Transformers, FastAPI, Python This project focused on building a Retrieval-Augmented Generation (RAG) chatbot to answer questions based on uploaded documents. The system was designed to assist users in querying large documents through natural language, without manually scanning content.

Key Components and Contributions:

• Developed a full RAG pipeline using LangChain, integrating document and semantic chunking, semantic embedding, vector store creation, and LLM inference.
• Implemented document embedding and retrieval using FAISS for fast similarity search.
• Designed a preprocessing module for semantic chunking to enhance retrieval accuracy.
• Integrated Hugging Face Transformers to generate context-aware answers based on retrieved chunks.
• Developed a user-friendly Streamlit interface for document upload and chatbot interaction.
• Built modular FastAPI endpoints for backend processing and model interaction.

Outcome: The system successfully demonstrated how users can obtain concise and relevant answers from large documents through an interactive UI. It also served as a template for future document-based assistants.