RAG Project

This project is a Retrieval-Augmented Generation (RAG) application with a Next.js client and a Node.js/Express server. It uses Redis (Valkey), Qdrant, and Google Gemini for Embeddings, and Ollama llema3 Model for document search and retrieval.

---

Project Structure

• client/ — Next.js frontend
• server/ — Node.js backend (Express, BullMQ, Qdrant, Ollama , llema3 Model)
• docker-compose.yml — For running Valkey (Redis-compatible)

---

Prerequisites

• Node.js (v18+ recommended)
• npm
• Docker (for Valkey/Redis)

---

Setup Instructions

1. Start Valkey (Redis-compatible)

docker-compose up -d

This will start Valkey on port 6379.

2. Server Setup

cd server
cp .env.example .env   # Fill in your API keys and Qdrant details
npm install
npm run dev            # Starts the Express API

Environment Variables (server/.env)

• GOOGLE_API_KEY — Your Google Gemini API key
• QDRANT_API_KEY — Qdrant API key
• QDRANT_URL — Qdrant endpoint (e.g., http://localhost:6333)
• QDRANT_COLLECTION — Qdrant collection name (default: rag_documents)
• REDIS_HOST — Redis/Valkey host (default: localhost)
• REDIS_PORT — Redis/Valkey port (default: 6379)

3. Client Setup

cd client
npm install
npm run dev            # Starts Next.js on http://localhost:3000

---

Usage

1. Upload PDF files via the client UI (handled by the server and processed into Qdrant)
2. Search and interact with your documents using the RAG-powered interface

---

Troubleshooting

• Ensure Valkey (Redis) is running on port 6379
• Ensure Qdrant is accessible and API keys are correct
• Check .env in the server folder for correct configuration

---

Scripts

Server

• npm run dev — Start Express server with hot reload
• npm run dev:worker — Start BullMQ worker for background PDF processing
• node search.js - To document search and retrieval with help of Ollama llama3 Model installed Locally

Client

• npm run dev — Start Next.js development server
• npm run build — Build for production
• npm start — Start production server

---

License

MIT