Smart India Hackathon 2025 (SIH25104)

An intelligent multilingual chatbot platform designed to democratize access to educational information across language barriers in India.

• Multilingual Support: English, Hindi, Marathi, and Marwari
• RAG-Powered Responses: Document-based question answering using FAISS vector search
• AI-Powered Chat: Google Gemini API for intelligent responses
• Admin Dashboard: Document management and system analytics
• Voice Input: Speech-to-text support (Google Cloud Speech)
• Handoff System: Escalate complex queries to human administrators

• Next.js 14 (App Router)
• TypeScript
• Tailwind CSS
• shadcn/ui components

• FastAPI (Python)
• Google Gemini API
• FAISS vector store
• Sentence Transformers
• Google Cloud Speech-to-Text

• Node.js 18+
• Python 3.9+
• Google Gemini API key

1. Clone repository

   git clone <repository-url>
   cd vidya-vaani-v0

2. Install frontend dependencies

   npm install

3. Install backend dependencies

   cd backend
   pip install -r requirements.txt
   cd ..

4. Configure environment

   Create .env.local in project root:

   NEXT_PUBLIC_DEV_MODE=false
   NEXT_PUBLIC_BACKEND_URL=http://localhost:8000
   GEMINI_API_KEY=your_gemini_api_key_here

   Create backend/.env:

   DEMO_MODE=false
   GEMINI_API_KEY=your_gemini_api_key_here

1. Start backend

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

2. Start frontend (in a new terminal)

   npm run dev

3. Access the application

   • Chat Interface: http://localhost:3000
   • Admin Dashboard: http://localhost:3000/admin

1. Navigate to Admin Dashboard
2. Go to "Documents" tab
3. Upload educational documents (PDF, DOCX, TXT, MD)
4. Documents are automatically indexed using FAISS

1. Select preferred language
2. Ask questions about uploaded documents
3. AI responds with context from the knowledge base

• Statistics: View system metrics and document counts
• Documents: Manage uploaded files
• Handoffs: Review queries escalated to human administrators

vidya-vaani-v0/
├── app/                    # Next.js app directory
│   ├── api/               # API routes
│   ├── admin/             # Admin dashboard pages
│   └── ...
├── backend/               # FastAPI backend
│   ├── app/
│   │   ├── api/          # API endpoints
│   │   ├── services/     # Business logic
│   │   └── core/         # Configuration
│   └── requirements.txt
├── components/            # React components
├── hooks/                 # Custom React hooks
├── lib/                   # Utility functions
├── retriever/            # RAG system (FAISS, embeddings)
├── public/               # Static assets
└── styles/               # Global styles

• POST /api/v1/chat/text - Send text message
• POST /api/v1/chat/audio - Send voice message

• POST /api/v1/documents/upload - Upload document
• GET /api/v1/documents/list - List all documents

• GET /api/v1/admin/stats - System statistics
• GET /api/v1/handoffs/list - List handoff requests

npm test                  # Frontend tests
cd backend && pytest      # Backend tests

npm run lint              # ESLint

Frontend (.env.local)

• NEXT_PUBLIC_DEV_MODE - Enable/disable dev mode (true/false)
• NEXT_PUBLIC_BACKEND_URL - Backend API URL
• GEMINI_API_KEY - Google Gemini API key

Backend (backend/.env)

• DEMO_MODE - Use mock services (true/false)
• GEMINI_API_KEY - Google Gemini API key
• SUPABASE_URL - Supabase URL (optional)
• SUPABASE_ANON_KEY - Supabase anon key (optional)

The system uses a modular service architecture with dependency injection:

1. Frontend - Next.js handles UI and API routing
2. Backend API - FastAPI orchestrates services
3. RAG Service - FAISS vector store for document retrieval
4. LLM Service - Google Gemini for response generation
5. Storage Service - Local file storage for documents

This project was developed for Smart India Hackathon 2025.

MIT License

• Google Gemini API for AI capabilities
• FAISS for efficient vector search
• Smart India Hackathon 2025 for the opportunity

Built with ❤️ for Smart India Hackathon 2025