Inspiration

As fitness enthusiasts, we found ourselves constantly switching between apps Strava for tracking, Instagram for sharing progress, Carousell for buying gear, and Eventbrite for finding local runs. We thought: why not one platform that connects all of these? Tribefit was born from the idea that fitness is inherently social and commercial you train with a community, you share milestones, and you buy and sell gear along the way.

What it does

Tribefit is a mobile-first fitness social platform that combines:

GPS activity tracking — Record runs, cycles, swims, and walks with live distance, pace, and speed using the Haversine formula: $$d = 2R \arcsin\left(\sqrt{\sin^2\left(\frac{\Delta\phi}{2}\right) + \cos(\phi_1)\cos(\phi_2)\sin^2\left(\frac{\Delta\lambda}{2}\right)}\right)$$ -Social feed — Share activities, like, comment, and follow other athletes
Marketplace — Buy and sell fitness gear with real-time chat between buyers and sellers
Groups & challenges — Join fitness communities, participate in merchant-hosted events and group challenges
Merchant ecosystem — Merchants list products, host events, and engage directly with the fitness community
AI food scanner — Snap a photo of any meal to receive an estimated nutritional breakdown, helping users make healthier food choices and monitor their daily intake.

How we built it

Frontend: React 18 + Vite, styled with custom CSS variables for dark/light theming
Backend: Supabase (PostgreSQL, Auth, Realtime, Row Level Security)
Maps: Leaflet + OpenStreetMap for live route tracking and activity maps
GPS: Custom useGPS hook with iOS-specific workarounds (polling fallback, relaxed accuracy thresholds, extended timeouts)
Real-time chat: Supabase Postgres Changes pub/sub on the chat_messages table with exponential backoff reconnection
Deployment: HTTPS dev server via @vitejs/plugin-basic-ssl for mobile GPS testing over local network

Challenges we ran into

iOS Safari GPS throttling — Safari stops delivering watchPosition callbacks after a few readings. We had to implement a 3-second getCurrentPosition polling fallback specifically for iOS devices.
HTTPS requirement — iOS blocks geolocation on non-HTTPS origins. Vite 5 removed the --https flag, so we needed to configure a plugin — and re-apply it every time a merge overwrote our config.
Supabase .single() crashes — Queries that return 0 rows throw "Cannot coerce the result to a single JSON object" instead of returning null. This caused silent failures on profile updates until we switched to .maybeSingle().
Mock vs. real data — The marketplace initially only displayed mock listings. Connecting to real Supabase data while keeping mocks as filler required careful tagging to prevent chat operations on non-existent records.
Team merge conflicts — Multiple teammates editing the same pages meant frequent conflict resolution while preserving each other's features.
Cross-feature integration — Because our app combines tracking, marketplace, nutrition and social networking, changes in one feature often affected another. Ensuring a smooth user experience across multiple interconnected systems required careful testing.

Accomplishments that we're proud of

Working GPS tracking on iPhone — Reliable distance tracking in a web app on iOS, with live map visualization and automatic fallback to manual input

End-to-end marketplace chat — Buyers can message sellers directly from a listing, with real-time message delivery via WebSockets
Role-based platform — Clean separation between Individual and Merchant experiences, enforced at both the UI and database level with Row Level Security
Zero external infrastructure costs — GPS via the browser Geolocation API, maps through OpenStreetMap, backend on Supabase's free tier and a lightweight frontend kept operating costs minimal.
Mobile-first responsive design — Works seamlessly on phones with bottom tab navigation, and adapts to desktop with a top nav

What we learned

The browser Geolocation API behaves very differently across platforms — you can't write it once and expect it to work everywhere
Supabase RLS is powerful but requires careful testing — a silent policy denial looks identical to "no data" from the client's perspective
Real-time features (chat, live tracking) require resilience patterns: reconnection backoff, missed message recovery, optimistic UI
Team collaboration with Git requires discipline — feature branches, atomic commits, and communication about which files you're touching

What's next for Tribefit

Route snapping via Mapbox Map Matching API to improve GPS distance accuracy by fitting raw points to real roads and trails
Push notifications for chat messages and event reminders
Activity photos — attach images to activities and feed posts
Leaderboards — weekly/monthly distance rankings within groups
Payment integration — in-app transactions for marketplace purchases
Native app wrapper (Capacitor/React Native) for better background GPS and notifications
Wearable integrations with Garmin, Apple Health and Google Fit for automatic activity synchronisation.
Smarter AI nutrition by expanding the existing food scanner into personalised meal recommendations based on activity history and fitness goals.