RAG chatbot (AIES project) 🤖💧

An intelligent chatbot designed to provide insights and answer queries about any pdf, this project leverages LLMs, vector databases to deliver accurate, contextual responses about information provided by the pdf.

Overview

This chatbot combines multiple advanced technologies to provide a comprehensive retrieval system:

• LLM-Powered Responses: Uses Google's Gemini 3.5 Flash Lite for intelligent query understanding and response generation
• Vector Search: Employs Chroma vector database with HuggingFace embeddings for semantic search over the information
• Web UI: FastAPI backend with a responsive React frontend for seamless interaction

Features

✨ Intelligent Query Processing

• Natural language query understanding
• Automatic SQL query generation for data retrieval
• Context-aware responses using both theoretical knowledge and real data

🗺️ Visual Data Representation

• Structured JSON responses
• Markdown support in the frontend

🚀 Modern Architecture

• FastAPI backend for high-performance API endpoints
• CORS-enabled for cross-origin requests
• Async request handling
• Structured output using Pydantic models

Tech Stack

Backend

• FastAPI - Modern Python web framework for building APIs
• LangChain - Framework for building LLM applications
• Google Generative AI - Gemini 3.5 Flash Lite LLM model
• Chroma - Vector database for semantic search
• HuggingFace - Pre-trained embeddings (intfloat/e5-large-v2)
• SQLAlchemy - ORM for database operations

Frontend

• React (running on localhost:5173)
• TypeScript/JavaScript

Installation

Prerequisites

• Python 3.8+
• Node.js (for frontend)
• Google API Key (for Gemini LLM)

Backend Setup

1. Clone the repository

2. Create a virtual environment

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install dependencies

   pip install -r requirements.txt
   

4. Set up environment variables Create a .env file in the root directory:

   GOOGLE_API_KEY=your_google_api_key_here
   

5. Run the backend server

   python main.py
   

   The API will be available at http://localhost:5000

Frontend Setup

1. Navigate to the frontend directory

   cd chatbotUI
   

2. Install dependencies

   npm install
   

3. Start the development server

   npm run dev
   

   The frontend will be available at http://localhost:5173

How It Works

Query Processing Pipeline

1. Input Processing: User query is received via FastAPI endpoint
2. Vector Search: Query is embedded and matched against documentation in Chroma vector store
3. Context Compilation: Theory from vector search
4. Response Generation: LLM generates a structured response with summary
5. JSON Response: Result is formatted and returned to frontend

Key Components

• Text Splitter: Chunks documents into 500-character segments with 70-character overlap for better retrieval
• Embedding Model: Uses intfloat/e5-large-v2 for semantic similarity
• LLM Model: Google's Gemini 2.5 Flash for fast and accurate responses

Configuration

Embedding Model

• Current: intfloat/e5-large-v2 (from HuggingFace)
• Customizable in main.py line 117

Chunk Parameters

• Chunk Size: 1000 characters
• Chunk Overlap: 70 characters
• Adjustable in main.py lines 105-108

Retriever Settings

• Search Type: Similarity search
• K (Top Results): 2 documents
• Modifiable in main.py line 177

Environment Variables

Create a .env file with the following variables:

GEMINI_API_KEY=your_api_key_here
# Add other environment variables as needed

Usage Example

1. Start both backend and frontend servers
2. Open the chatbot UI at http://localhost:5173
3. Start asking questions

Contributing

Contributions are welcome! Please feel free to submit issues or pull requests.

License

This project is built for the Smart India Hackathon. Check the repository for specific license details.

Contact & Support

For issues, questions, or suggestions, please create an issue in the repository.

Acknowledgments

• Google Cloud for Gemini API
• HuggingFace for embeddings
• LangChain community for excellent LLM framework