Autriv is a next-gen fitness and nutrition app built to replace the friction of manual tracking with intelligence. I built this to solve a simple problem: manual logging is slow and most workout apps don't evolve with you.
By combining GPT-4o Vision with a Serverless Edge architecture, Autriv turns a photo of your meal into macro data and builds workout protocols that adapt based on how you actually perform.
Autriv runs on a modern, decoupled stack built for speed:
- Frontend: React Native (Expo) with a reactive engine for 60fps performance.
- Backend: Supabase (Postgres + Auth + Realtime).
- AI Logic: Supabase Edge Functions routing to OpenAI’s multimodal models.
- Data: Verified cross-referencing via OpenFoodFacts.
No more searching for "chicken breast" in a database of 10,000 entries.
- Vision: Snap a photo. The app uses GPT-4o to identify ingredients, estimate portions, and calculate macros.
- Voice/Text: Just say "I had a protein shake and a banana," and the app parses it into structural data.
- Verification: It double-checks AI estimates against real food databases to keep the numbers accurate.
A workout engine that actually understands progressive overload.
- Dynamic Plans: Programs like "Anabolic Foundation" that track your volume and intensity.
- On-the-Fly Generator: If you're short on time or equipment, the app builds a custom session instantly.
- Offline-First: Built with React Context and AsyncStorage so it works in basement gyms with zero signal.
If a fitness app feels laggy, it’s useless.
- Skia Graphics: High-performance, GPU-accelerated charts for weight and macro trends.
- Smooth Animations: 60fps transitions using
react-native-reanimated.
| Layer | Technology |
|---|---|
| Mobile | React Native / Expo (SDK 54) |
| Language | TypeScript |
| Database | PostgreSQL (Supabase) |
| Logic | Supabase Edge Functions (Deno) |
| AI | OpenAI GPT-4o Vision & Chat |
| UI | Skia, Reanimated, React Native Paper |
- Shipping: Managed the full lifecycle from an idea to a Live App Store Deployment.
- AI Integration: Built the prompt chains and vision pipelines to handle unstructured data.
- System Design: Designed a scalable, serverless backend that handles heavy AI tasks at the edge.
- Performance: Optimized the app for fluid UI/UX even when handling large biometric datasets.
Built by Jackson Hemopo