RAGPickerz

RAGPickerz is a Retrieval-Augmented Generation (RAG) system designed for document analysis and question answering. It was developed for the Bajaj Finserv Hackathon.

Features

• Document Loading: Supports PDF and DOCX file formats
• Vector Store: Uses FAISS for efficient document embeddings and similarity search
• Semantic Chunking: Splits documents into meaningful chunks using semantic analysis
• Question Answering: Processes questions and retrieves relevant answers from documents
• Batch Processing: Handles multiple questions concurrently
• FastAPI Integration: RESTful API for easy integration

Installation

1. Clone the repository:

git clone https://github.com/iamanimeshdev/RAGPickerz.git
cd RAGPickerz

2. Install dependencies:

pip install -r requirements.txt

3. Set up environment variables:

cp .env.example .env
# Edit .env with your configuration

Usage

Running the API Server

# Start the FastAPI server
uvicorn app.main:app --reload

The API will be available at http://localhost:8000

Making API Requests

Endpoint: /api/v1/hackrx/run

Method: POST Content-Type: multipart/form-data

Parameters:

• questions: List of questions (JSON array)
• file: Document file (PDF or DOCX)

Example Request:

curl -X POST "http://localhost:8000/api/v1/hackrx/run" \
  -F "questions=[\"What is the policy coverage?\", \"What are the exclusions?\"]" \
  -F "file=@document.pdf"

Response:

{
  "answers": [
    "The policy covers medical expenses up to 5 lakhs.",
    "Pre-existing conditions are excluded from coverage."
  ]
}

Running the Pipeline Directly

python rag_pipeline/main.py

Configuration

The system uses the following configuration:

• Vector Database Path: .\faiss_index
• Chunk Size: 350 characters
• Chunk Overlap: 50 characters
• Embedding Model: sentence-transformers/all-MiniLM-L6-v2
• LLM Model: gemma-3n-e2b-it

File Structure

RAGPickerz/
├── app/
│   ├── __init__.py
│   ├── main.py              # FastAPI application entry point
│   └── routers/
│       └── hackrx.py        # API router for HackRx endpoints
├── faiss_index/             # Directory for FAISS vector store
├── rag_pipeline/
│   ├── __init__.py
│   ├── config.py           # Configuration and model setup
│   ├── document_loader.py  # Document loading utilities
│   ├── embedder.py         # Vector store building and management
│   ├── query_pipeline.py   # Query processing pipeline
│   ├── retriever.py        # Document retrieval from vector store
│   └── main.py             # Main pipeline execution
├── requirements.txt         # Python dependencies
├── readme.md               # This file
└── .gitignore              # Git ignore rules

Supported Document Formats

• PDF (.pdf)
• Word Documents (.docx)

Performance

• Document Loading: Optimized for large documents
• Vector Indexing: FAISS provides fast similarity search
• Concurrent Processing: Handles multiple questions simultaneously
• Caching: Vector store is persisted to disk for reuse

Error Handling

The system includes comprehensive error handling:

• Unsupported file types
• Empty documents
• Rate limiting for API requests
• Server errors with descriptive messages

Development

Prerequisites

• Python 3.8+
• pip
• A Google AI Platform account (for LLM access)

Environment Variables

Create a .env file with the following:

# Google AI Platform credentials
GOOGLE_API_KEY=your_api_key_here

Testing

# Run the test suite
python -m pytest tests/

License

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

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

Support

For issues and questions, please open an issue on the GitHub repository.