REVA AI — Agentic Complaint Portal

Overview

A secure, multilingual, AI-agent-driven complaint intake system with:

• Citizen Portal: Voice-first complaint filing with AI agent guidance
• Police Portal: Multi-station dashboard with RBAC and data isolation
• Backend: Express + Prisma + PostgreSQL API gateway
• AI Integration: Direct Azure OpenAI + Azure Speech integration in frontend

---

📁 Project Structure

proto-ag/
├── frontend/          # React + Vite (Citizen & Police UI)
├── backend/           # Express.js API Gateway + Prisma ORM
│   ├── src/
│   │   ├── routes/    # auth, complaints, police, stations, analytics
│   │   ├── middleware/ # auth, audit, errorHandler
│   │   └── utils/     # prisma, logger
│   └── prisma/        # schema.prisma
└── docker-compose.yml

---

🚀 Quick Start

Prerequisites

• Node.js 18+
• Python 3.11+
• PostgreSQL (or use Docker)

1. Backend

cd backend
cp .env.example .env
# Edit .env with your database and Azure credentials
npm install
npx prisma migrate dev
npm run dev

3. Frontend

cd frontend
npm install
npm run dev

---

🐳 Docker Deployment

docker-compose up -d

---

🔑 Environment Variables

Backend (backend/.env)

Variable	Description
DATABASE_URL	PostgreSQL connection string
JWT_ACCESS_SECRET	Access token signing secret
JWT_REFRESH_SECRET	Refresh token signing secret
AZURE_OPENAI_ENDPOINT	Azure OpenAI endpoint
AZURE_OPENAI_KEY	Azure AI services API key
AZURE_SPEECH_KEY	Azure Speech Services key
AI_SERVICE_URL	Python AI microservice URL

AI Service (ai-service/.env)

Variable	Description
AZURE_OPENAI_ENDPOINT	Azure OpenAI endpoint
AZURE_OPENAI_KEY	API key
AZURE_OPENAI_DEPLOYMENT	Deployment name (sih-vision)
AZURE_SPEECH_KEY	Speech service key
AZURE_SPEECH_REGION	eastus2

---

🏛️ Architecture

React (port 3000)
    ↓        ↘
Express API    Azure OpenAI +
Gateway        Azure Speech
(port 5000)
    ↓
PostgreSQL
(Prisma ORM)

---

🤖 AI Agent Stages

1. Safety Check — Is user safe?
2. Open Narrative — Free-form story collection
3. Structured Extraction — Field-by-field extraction
4. Missing Data Loop — Targeted follow-ups (max 5)
5. Threat Detection — Risk scoring 0-100
6. Summary Generation — FIR-style formal summary
7. User Confirmation — Final verification

---

👮 Police RBAC

Role	Capabilities
SUPER_ADMIN	All stations, audit, analytics, configure
STATION_ADMIN	Manage officers, assign tasks, configure geofence
OFFICER	View/update assigned complaints only

---

🔒 Security

• Row-level security via Prisma query scoping
• JWT access (15m) + refresh (7d) token rotation
• Refresh token reuse detection with revocation
• HTTP-only secure cookies
• Rate limiting (100 req/15min, 10 for OTP endpoints)
• IP logging + audit trail for all mutations
• Aadhaar masked before storage (XXXX-XXXX-XXXX)

---

📞 Emergency Contacts

The system automatically detects emergency keywords and displays:

• 📞 112 — National Emergency
• 📞 100 — Police
• 📞 1091 — Women helpline
• 📞 1098 — Child helpline

---

🌐 Supported Languages

Hindi, Tamil, Telugu, Kannada, Marathi, Bengali, Gujarati, Punjabi, Odia, Malayalam, English