KOM Hunter

The Question

Why is finding segments worth hunting still a game of scroll-and-hope?

I ride Cape Town road and trail. I've spent hours on Strava looking for the right segment for a given session — short enough to red-line, long enough to train against, within reach of where I actually ride. The heatmap tells you where people go. The leaderboard tells you who was fastest. Nothing tells you which segments are achievable for you.

What I Built

A Strava companion that surfaces segments through five lenses. Around Me for discovery near a pin. Your Speed for segments where your current fitness predicts you could place. Hidden Gems for quiet segments with thin leaderboards. Legendary for the high-prestige KOMs worth the pilgrimage. On This Route for segments hiding along a planned ride.

The backend ingests the full Strava corpus for every connected athlete, extracts polylines from activity streams, classifies surface from heading and terrain, tracks your trophy history per segment, and runs an achievability prediction on every result. React Native and Expo front-end. Supabase backend. Mapbox for the map.

What I Learned

Strava's API is a trap door full of quiet surprises. Endpoints silently return empty objects. Docs lie about which query parameters are honoured. Rate limits hit you mid-sync with no retry etiquette. Shipping this has been as much about building an observability layer for someone else's API as it has been about building the app itself.

Status

MVP ready at komhunter.xyz. Polyline coverage growing daily. Awaiting Strava app approval before beta onboarding.