📚 RAG Chatbot

Your Documents. Your Questions. Instant AI-Powered Answers.

Quick Start • Features • Architecture • API Docs

---

🌟 What is RAG Chatbot?

RAG Chatbot is a cutting-edge Retrieval-Augmented Generation (RAG) application that transforms how you interact with your documents. Upload any PDF, and engage in intelligent, context-aware conversations powered by Google's Gemini AI.

Unlike traditional chatbots, RAG Chatbot doesn't hallucinate — it answers based strictly on the content of your uploaded documents, combining the power of semantic search with advanced language models.

🎯 Why RAG Chatbot?

• ✅ 100% Context-Grounded: Answers derived exclusively from your documents
• ⚡ Lightning Fast: Optimized retrieval with FAISS vector database
• 🎨 Premium UI/UX: Glassmorphism design with smooth animations
• 🔒 Privacy-First: Process documents locally with no data persistence on restart

---

🎥 Demo

---

🚀 Quick Start

Get up and running in 3 minutes!

Prerequisites

• Python 3.9+ — Download
• Node.js 18+ — Download
• Google API Key — Get one here

Installation

1️⃣ Clone the Repository

git clone https://github.com/yugam23/RAG-Chatbot.git
cd RAG-Chatbot

2️⃣ Backend Setup

cd backend
python -m venv venv

# Windows:
venv\Scripts\activate
# macOS/Linux:
source venv/bin/activate

pip install -r requirements.txt

Configure Environment: Create a .env file in backend/:

GOOGLE_API_KEY=your_actual_api_key_here
ALLOWED_ORIGINS=http://localhost:5173

Run the Backend:

python main.py
# 🚀 Server running at http://localhost:8000

3️⃣ Frontend Setup

Open a new terminal:

cd frontend
npm install
npm run dev
# ✨ App running at http://localhost:5173

---

✨ Key Features

🧠 Intelligent Backend

Feature	Description
🤖 Google Gemini Integration	Powered by gemini-flash-latest for ultra-fast, accurate responses
🔍 Advanced RAG Pipeline	Semantic chunking (800 chars, 400 overlap) + Gecko embeddings
💾 Vector Search	FAISS CPU-optimized indexing with k=7 retrieval
📑 Robust PDF Processing	Magic byte validation, 50MB limit, secure temp storage
💬 Session Management	SQLite-based chat history with full persistence
🔄 Auto-Reset	Session and index auto-clear on server restart

🎨 Premium Frontend

Feature	Description
✨ Glassmorphism Design	Modern blur effects, gradients, and depth
🎬 Startup Animation	Smooth logo intro with motion transitions
💬 Real-Time Streaming	Server-Sent Events (SSE) for live response rendering
📝 Markdown Support	Full syntax highlighting with react-markdown + remark-gfm
📱 Fully Responsive	Optimized for desktop, tablet, and mobile
⚡ Optimized Caching	TanStack Query for efficient data fetching

🏗️ Architecture

RAG Chatbot follows a modern client-server architecture:

┌─────────────────────────────────────────────────────────────────┐
│                         USER INTERFACE                          │
│  React + Vite + Tailwind CSS + Framer Motion + TanStack Query   │
└────────────────────────────┬────────────────────────────────────┘
                             │ HTTP/SSE
                             ▼
┌─────────────────────────────────────────────────────────────────┐
│                        FASTAPI BACKEND                          │
│  ┌──────────────┐  ┌──────────────┐  ┌─────────────────────┐    │
│  │   Routers    │  │  Middleware  │  │   State Manager     │    │
│  │ /upload      │  │ - Rate Limit │  │ - Session State     │    │
│  │ /chat        │  │ - Request ID │  │ - Vector Store Ref  │    │
│  └──────────────┘  └──────────────┘  └─────────────────────┘    │
│                             │                                   │
│                             ▼                                   │
│  ┌──────────────────────────────────────────────────────────┐   │
│  │                  RAG PIPELINE (LangChain)                │   │
│  │  ┌────────┐   ┌────────┐   ┌──────────┐   ┌──────────┐   │   │
│  │  │  PDF   │──▶│ Chunk  │──▶│  Embed   │──▶│  FAISS │   │   │
│  │  │ Loader │   │ (800)  │   │ (Gecko)  │   │  Index   │   │   │
│  │  └────────┘   └────────┘   └──────────┘   └──────────┘   │   │
│  └──────────────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────────────┘

🔄 RAG Pipeline Workflow

1. 📤 Document Ingestion: PDF Upload → Magic Byte Validation → Text Extraction → Recursive Chunking (800/400).
2. 🧮 Embedding & Indexing: Google Gecko Embeddings → FAISS Vector Store.
3. 💬 Query Processing: User Query → Embedding → Similarity Search (Top-k) → Prompt Construction.
4. 🤖 Response Generation: Context + Query → Gemini Flash → Streaming Response.

📂 Project Structure

RAG-Chatbot/
├── 📁 backend/                     # FastAPI Python Backend
│   ├── 📁 routers/                 # API Route Handlers
│   │   ├── upload.py               # PDF upload & indexing endpoint
│   │   └── chat.py                 # Chat streaming & history endpoints
│   ├── 📁 tests/                   # Pytest Test Suite
│   ├── 📁 temp/                    # Temporary PDF storage
│   ├── 📁 faiss_index/             # Vector database (generated)
│   ├── config.py                   # Centralized configuration
│   ├── database.py                 # SQLite async operations
│   ├── ingestion.py                # Document processing pipeline
│   ├── rag.py                      # RAG chain implementation
│   ├── state.py                    # Application state management
│   ├── vector_store.py             # Vector store logic
│   ├── middleware.py               # Rate limiting & request tracking
│   ├── logging_config.py           # Structured logging setup
│   ├── models.py                   # Pydantic models
│   ├── main.py                     # FastAPI app entry point
│   └── requirements.txt            # Python dependencies
│
├── 📁 frontend/                    # React + Vite Frontend
│   ├── 📁 src/
│   │   ├── 📁 components/          # React Components (Header, ChatArea, etc.)
│   │   ├── 📁 hooks/               # Custom React Hooks (useChat, useApiQueries)
│   │   ├── 📁 services/            # API Communication
│   │   ├── 📁 context/             # React Context Providers
│   │   ├── App.jsx                 # Main App component
│   │   ├── main.jsx                # React entry point
│   │   └── index.css               # Global styles & theme
│   ├── 📁 public/                  # Static Assets
│   └── vite.config.js              # Vite configuration
│
└── 📄 README.md                    # This file!

⚙️ Configuration

Environment Variables

Variable	Required	Default	Description
GOOGLE_API_KEY	✅ Yes	—	Google AI API key for Gemini & Gecko
ALLOWED_ORIGINS	❌ No	http://localhost:5173	Comma-separated CORS origins

Advanced Tuning (backend/config.py)

# Document Processing
CHUNK_SIZE = 800              # Characters per chunk
CHUNK_OVERLAP = 400           # Overlap between chunks

# Retrieval
RETRIEVER_K = 7               # Number of chunks to retrieve

# Models
EMBEDDING_MODEL = "models/text-embedding-004"
LLM_MODEL = "gemini-flash-latest"

📡 API Reference

Full interactive documentation available at http://localhost:8000/docs.

Core Endpoints

• POST /upload: Upload and index a PDF. Validates magic bytes and size.
• POST /chat: Stream chat response via SSE. Requires active session.
• GET /history: Retrieve stored chat history.
• POST /clear-chat: Clear history but keep document index.
• POST /reset: Full session reset (wipes history + index).

🧪 Testing & Troubleshooting

Running Tests

cd backend
pytest -v          # Run all tests
pytest tests/test_ingestion.py  # Test specific module

Common Issues

• ValidationError: GOOGLE_API_KEY field required: Add your API key to backend/.env.
• Failed to fetch: Ensure backend is running on port 8000.
• Invalid file type: Ensure the file is a valid PDF.

---

Made with ❤️ by Yugam

⬆ Back to Top