Skip to content

Readme.md

Latest commit

 

History

History
252 lines (184 loc) · 6.54 KB

Readme.md

File metadata and controls

252 lines (184 loc) · 6.54 KB

QueryWise Logo

🚀 QueryWise – AI Powered Document Query System

🔗 Live Demo: 🌐 Live Demo

AI-Powered Document Query System – Upload PDFs or share document URLs and ask natural language questions to get instant, intelligent answers.
Built with a cutting-edge stack: FastAPI (backend) ⚡ + Next.js (frontend) 🎨.

Contributions Welcome GitHub stars License Next.js FastAPI Python

Key Features

  • Document Processing – Upload PDFs or provide document URLs for seamless ingestion
  • Google Drive Smart Links – Auto-converts shareable Drive links into direct download links
  • AI-Powered Q&A – Powered by Google Gemini 2.0 Flash Lite for intelligent responses
  • Vector Search – Fast & accurate retrieval using FAISS + Jina embeddings
  • Persistent Caching – Indexes cached with SHA256 hashing for lightning-fast reuse
  • Modern UI/UX – Clean, responsive design with dark & light mode support
  • Real-time Processing – Concurrent question handling for speed & scalability
  • Secure by Design – Bearer token authentication & input validation

Why QueryWise?

✔️ No more manual searching through huge PDFs
✔️ Get answers with context in seconds
✔️ Works for research, legal docs, policies, contracts, study notes
✔️ Fast ⚡, Secure 🔒, and Beautiful ✨

Sneak Peek (UI)

Light Mode 🌞 Dark Mode 🌙
Light UI Dark UI

Additional Screens:
Options
Contact

Architecture

Backend (FastAPI)

  • Framework: FastAPI with async support
  • Document Processing: PyMuPDF for PDF parsing
  • Vector Store: FAISS with Jina embeddings
  • LLM: Google Generative AI (Gemini 2.0 Flash Lite)
  • Text Splitting: LangChain's RecursiveCharacterTextSplitter
  • Caching: File-based persistent indexes with SHA256 hashing

Frontend (Next.js)

  • Framework: Next.js 15 with TypeScript
  • UI Components: Radix UI with Tailwind CSS
  • State Management: React hooks
  • Theme Support: next-themes for dark/light mode
  • File Handling: Native FormData API for file uploads

Prerequisites

  • Python 3.8+
  • Node.js 18+
  • npm or yarn
  • API Keys:
    • Google AI API key (for Gemini)
    • Jina AI API key (for embeddings)

Installation

Backend Setup

  1. Navigate to the backend directory:
cd backend
  1. Install Python dependencies:
pip install -r requirements.txt
  1. Create a .env file with your API keys:
GOOGLE_API_KEY=your_google_api_key_here
JINA_API_KEY=your_jina_api_key_here
BEARER_TOKEN=your_secure_bearer_token_here
  1. Start the FastAPI server:
uvicorn main:app --reload --host 0.0.0.0 --port 8000
  1. Production ready deployment:
current setup uses Hugging Face Spaces (Docker build)

Frontend Setup

  1. Navigate to the frontend directory:
cd frontend
  1. Install Node.js dependencies:
npm install
  1. Create a .env.local file:
NEXT_PUBLIC_API_URL=http://localhost:8000/hackrx/run
NEXT_PUBLIC_API_TOKEN=your_secure_bearer_token_here
  1. Start the development server:
npm run dev

The application will be available at http://localhost:9002.

  1. Production ready deployment:
current setup uses Vercel.

API Endpoints

POST /hackrx/run

Process documents and answer questions.

Request Body (multipart/form-data):

  • questions: Array of strings (questions to ask)
  • file: PDF file (optional)
  • documents: Document URL (optional)

Response:

{
  "answers": ["Answer 1", "Answer 2", "..."]
}

Authentication: Bearer token required in Authorization header.

GET /

Health check endpoint.

Usage

  1. Document Input: Choose between uploading a PDF file or providing a document URL (including Google Drive links)
  2. Questions: Enter your questions, one per line
  3. Processing: The system will:
    • Download/process the document
    • Create vector embeddings (cached for future use)
    • Use RAG (Retrieval-Augmented Generation) to answer questions
  4. Results: View AI-generated answers with source context

Key Technologies

  • LangChain: Document processing and RAG pipeline
  • FAISS: Vector similarity search
  • Jina Embeddings: High-quality text embeddings
  • Google Gemini: Large language model for answer generation
  • Radix UI: Accessible component library
  • Tailwind CSS: Utility-first CSS framework

Performance Features

  • Concurrent Processing: Multiple questions processed simultaneously
  • Intelligent Caching: Document indexes persisted to disk
  • Chunking Strategy: Optimized text splitting (1000 chars, 150 overlap)
  • Retrieval Optimization: Top-7 relevant chunks for context

Security

  • Bearer token authentication
  • Input validation and sanitization
  • CORS configuration for frontend communication
  • Secure file handling with temporary files

Development

Backend Development

cd backend
uvicorn main:app --reload --host 0.0.0.0 --port 8000

Frontend Development

cd frontend
npm run dev

Type Checking (Frontend)

npm run typecheck

Use Cases

  • Students: Quickly summarize large PDFs and extract key points
  • Enterprises: Analyze contracts, policies, and reports
  • Legal Professionals: Retrieve specific clauses from long agreements
  • Developers: Integrate smart document search into apps

Contributing

  1. Fork the repository
  2. Create a feature branch
  3. Make your changes
  4. Optimize caching or embedding logic
  5. Add tests if applicable
  6. Submit a pull request

License

This project is developed by team Innov8 as a solution to HackRX6.0 Hackathon by Bajaj Finserv.

Support

For questions or issues, please refer to the application's built-in FAQ section or contact the development team.

Made with ❤️ by Team Innov8

⚡