Building the MVP in 48 Hours: Architecture, Stack, and Lessons
To build a working prototype in such a short time, we needed a flexible and scalable stack. Our goal: to create a minimal yet functional MVP that could grow over time.
Our Tech Stack
- Frontend: Flutter + Riverpod + go_router
- Backend: Node.js + PostgreSQL with PostGIS for geospatial queries
- Image Storage: DigitalOcean Spaces
- Authentication and Notifications: Firebase Auth + FCM
We followed the Clean Architecture pattern to ensure clear separation between presentation, domain, and data layers. Even at MVP stage, we aimed for scalable code.
Core Features in the MVP
- Report a lost pet with location and images
- Submit a sighting (anonymous or logged in)
- View reports on an interactive map
- Push notifications for nearby sightings
- Pet registration with breed, color, size, etc.
Main Technical Challenges
- Accurate geolocation: using GPS and manual input, stored as spatial points in the database.
- Image handling: image compression + CDN delivery for cost-effective performance.
- Matching: we prepared the backend for future AI image-matching, but included filters by species, color, size, and location.
- UX: it had to be intuitive — one minute to report, even for non-technical users.