📄 AI Knowledge Base Chatbot

Overview

This project is a Retrieval-Augmented Generation (RAG) AI chatbot that can answer questions over uploaded PDFs.

• Upload multiple PDFs
• Search across documents
• Get answers grounded in the documents
• Track conversation history per PDF
• See sources for each answer

Features

• ✅ Upload PDFs and manage multiple documents
• ✅ Ask questions about uploaded PDFs
• ✅ Multi-PDF chat memory
• ✅ Source citations for each answer
• ✅ Fully deployable with Docker

Tech Stack

• LLM: Ollama LLaMA 3
• Framework: LangChain
• Embeddings: HuggingFace Sentence Transformers
• Vector DB: Chroma
• Backend: FastAPI
• Frontend: Streamlit
• Deployment: Docker

Project Structure

rag-chatbot/
│
├── backend/
│ ├── backend.py
│ ├── requirements.txt
│
├── frontend/
│ ├── frontend.py
│ ├── requirements.txt
│
├── uploads/ # Folder for uploaded PDFs (auto-created)
├── backend.log # Generated by logging (auto-created)
├── Dockerfile
└── README.md

Setup Locally

Backend

cd backend
pip install -r requirements.txt
uvicorn backend:app --reload

Frontend

cd frontend
pip install -r requirements.txt
streamlit run frontend.py

Docker

docker build -t rag-chatbot .
docker run -p 8000:8000 -p 8501:8501 rag-chatbot

Usage

1. Open the frontend in your browser
2. Upload a PDF (or multiple PDFs)
3. Select a PDF from the sidebar
4. Ask questions; answers appear with chat history visible
5. Switch PDFs anytime using the sidebar
6. See sources for each answer

Note: Chat history is saved per PDF while the session is running. Backend logs are saved in backend.log.

---

Demo: