🧠 AI Document Search (RAG Chatbot)

A Retrieval-Augmented Generation (RAG) chatbot that lets you search and query documents (web pages, PDFs, etc.) using a Natural Language interface — built with FastAPI + LangChain (backend) and React + TypeScript (frontend).

---

📂 Project Structure

AI_Document_Search_RAG_Chatbot/
│
├── rag_chatbot/ # Backend (FastAPI + LangChain + Chroma)
│ ├── config/ # Configuration & env variables
│ ├── preprocessing/ # Chunking, embeddings, vector store creation
│ ├── utilities/ # Document loaders & retrievers
│ ├── main.py # FastAPI app entrypoint
│ ├── requirements.txt
│ ├── README.md (optional backend-specific)
│
├── rag-chatbot-frontend/ # Frontend (React + TypeScript)
│ ├── src/api/ # API client
│ ├── src/components/ # UI Components (Chat UI)
│ ├── public/
│ ├── package.json
│ ├── tsconfig.json
│ ├── README.md (optional frontend-specific)
│
└── .env.example # Template env file

---

⚙️ Features

• Document Loading: Load from URLs or PDF files.
• Preprocessing: Text chunking & token-aware splitting.
• Vector Search: Embed & store in Chroma vector DB.
• LLM Integration: Works with Perplexity.ai API (Sonar model) or OpenAI.
• Frontend UI: Simple chat interface to query your document.
• MLOps Ready: Modular structure, .env config, API-key safety, vector DB persistence.

---

🚀 Getting Started

1️⃣ Clone the Repository

git clone https://github.com/sagarmandiya/AI_Document_Search_RAG_Chatbot.git

cd "AI_Document_Search_RAG_Chatbot"

---

🔹 Backend Setup (FastAPI + LangChain)

Prerequisites

• Python 3.10+ and pip
• virtualenv (optional but recommended)

Install dependencies

python -m venv .venv  
source .venv/bin/activate # Windows: .venv\Scripts\activate  
pip install -r requirements.txt  

Environment Variables

Create a .env in rag_chatbot/:

# Document source to index

DOC_SOURCE_URL=https://lilianweng.github.io/posts/2023-06-23-agent/

# Vector DB location

VECTOR_DB_DIR=./vector_store

# Models

EMBED_MODEL=all-MiniLM-L6-v2
LLM_MODEL=sonar

# LangChain API tracing (optional)

LANGCHAIN_TRACING_V2=true
LANGCHAIN_ENDPOINT=https://api.smith.langchain.com
LANGCHAIN_API_KEY=your-langchain-api-key

# Perplexity API Key

PPLX_API_KEY=your-perplexity-api-key

⚠️ Do NOT commit .env to GitHub — keep it in .gitignore.

Run the Backend

uvicorn main:app --reload

Server runs at: http://127.0.0.1:8000

---

🔹 Frontend Setup (React + TypeScript)

Install Dependencies

cd ../rag-chatbot-frontend  
npm install

API Configuration

Edit src/api/chatbot.ts to point to your backend URL:

const API_BASE_URL = "http://127.0.0.1:8000";

If backend runs locally, you can also add in package.json:

"proxy": "http://127.0.0.1:8000"

Run Frontend

npm start

Frontend runs at: http://localhost:3000

---

📡 Example Usage

Once both frontend and backend are running:

1. Open http://localhost:3000 in your browser
2. Type: What is Task Decomposition?
3. Bot responds with context‑aware answer from the indexed document(s)

---

🛠 MLOps & Deployment Notes

• Persistence: Vector DB stored locally in vector_store/. Deploy to cloud storage for production.
• Keys: Store API keys in environment variables — never commit to GitHub.
• Monitoring: Integrate logging / tracing via LangChain's tracing API.
• Scaling: Deploy backend to services like Render, AWS Lambda, Azure Functions and frontend to Netlify/Vercel.
• Testing: Add pytest for backend and React Testing Library/Jest for frontend.

---

📦 Requirements

Backend (rag_chatbot/requirements.txt)

langchain
langchain-community
langchain-core
bs4
pypdf
sentence-transformers
chromadb
fastapi
uvicorn
python-dotenv

Frontend (rag-chatbot-frontend/package.json - dependencies excerpt)

"dependencies": {
"axios": "^latest",
"react": "^18.x",
"react-dom": "^18.x",
"typescript": "^5.x",
"@types/react": "^18.x",
"@types/react-dom": "^18.x"
}

---

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

🙌 Acknowledgements

• LangChain
• FastAPI
• React
• Chroma
• Perplexity