RAG-Powered News Chatbot

A full-stack chatbot that answers queries over recent news using Retrieval-Augmented Generation (RAG), powered by Jina embeddings, Qdrant vector search, and Gemini/OpenAI LLMs.

---

Demo Video

Link - https://drive.google.com/file/d/1Z2yCFecZvgwIIs2IYPhgSNtRgxY3Yt2k/view?usp=drive_link

---

Features

• RAG pipeline with top-k context retrieval
• Gemini or OpenAI GPT for answering questions
• Vector search using Qdrant
• Chat interface with session history
• Redis-backed memory per session
• Responsive UI built with React + Tailwind CSS

---

Tech Stack

• Frontend: React + Vite + Tailwind CSS
• Backend: Node.js + Express
• Vector DB: Qdrant
• Embedding: Jina AI
• LLM: Gemini (or OpenAI GPT)
• Storage: Redis Cloud (session data)

---

Setup Instructions

1. Clone & Install

git clone https://github.com/rekha0suthar/rag-chatbot.git
cd rag-chatbot

2. Environment Variables

Backend .env:

PORT=5000
REDIS_URL=redis://default:<password>@<host>:<port>
QDRANT_URL=http://localhost:6333
JINA_API_KEY=your_jina_api_key
GEMINI_API_KEY=your_makersuite_key  # OR use OPENAI_API_KEY

Frontend:

API_BASE=https://rag-chatbot-server.vercel.app

3. Run

# backend
cd backend
npm install
node index.js

# frontend
cd frontend
npm install
npm run dev

---

Deployment

• Frontend & Backend: Deploy to Vercel
• Redis: Use Redis Cloud free tier
• Qdrant: Use Qdrant Cloud or Docker