Ride-hailing MVP — driver Android app + passenger web app + admin panel (Node/React Native)

**Description:** We're building a city taxi booking platform for a single city in North India. We need a working MVP, not a white-label customisation. The codebase must be fully owned by us. **What we need built** 1. Backend API — Node.js + TypeScript, PostgreSQL with PostGIS, Redis. Driver/vehicle/ride/subscription models, OTP auth, document records with verification audit trail. 2. Dispatch service — real-time matching of nearest available driver, configurable acceptance window with auto-reassign on timeout or rejection, WebSocket location streaming. 3. Driver app — Android only, React Native or Flutter. Onboarding with document upload, online/offline toggle, ride accept/reject, in-ride flow, earnings screen, subscription balance, mandatory panic button, handoff to Google Maps for navigation. 4. Passenger web app — mobile-first PWA, installable. Opens from a QR code with no app install required. Location entry, fare estimate, booking, live tracking, panic button, ride history. 5. Admin panel — driver approval queue with verification logging, live fleet map, ride and dispatch logs, subscription ledger, fare configuration, and a restricted operator role for a 24x7 control room with live emergency alerts. **Hard requirements** - Three languages selected on first run and switchable: English, Hindi, Punjabi. All strings externalised — no hardcoded text. - Routing, distance and ETA via self-hosted OSRM. Google Maps Directions API must not be used for routing. Map display via MapLibre with open tiles. - Background location must survive aggressive battery managers on Xiaomi, Vivo, Oppo and Realme devices. Foreground service required. - The driver app must behave sensibly through network loss and reconnect mid-ride. Queued location writes, no lost ride state. - Live location tracking and panic button are regulatory requirements, not optional features. Build them into the schema from the start. - Call masking integration (Exotel or similar) so phone numbers are never exposed. - All data stored on Indian-region infrastructure. **Out of scope** No iOS apps. No in-app fare payments — fares settle directly between passenger and driver. No wallet, chat, ratings, promo codes, surge pricing, scheduled rides, or multiple vehicle categories. **Deliverables** Source code in our GitHub repository from day one with regular pushes. Deployment documentation, database schema documentation, API documentation, and a signed Play Store build. Sixty days of bug-fix support after handover. **Milestones** 1. Database schema and backend API 2. Dispatch service, demonstrated working with simulated drivers before any UI work 3. Admin panel 4. Driver Android app 5. Passenger PWA 6. Integration, testing, Play Store submission Payment is per milestone on acceptance. No large upfront payment. **Please answer these in your proposal — generic proposals will be ignored** 1. How do you keep background location alive on a Xiaomi or Vivo phone when the app is not in the foreground? Be specific. 2. Have you deployed OSRM or another self-hosted routing engine? Which project? 3. A driver loses network for four minutes mid-ride in a hilly area. What happens to the ride state and the location history? 4. Link to a live app you have built with real-time location tracking. App store links or a demo we can run. Also tell us your availability per week and your realistic timeline. We will share a clickable prototype and a detailed functional specification with shortlisted candidates after an NDA.