AI Golf Swing
A golf-performance web app: capture a swing, get AI analysis with a 3D avatar and skeletal metrics, practise drills and get reviewed by a coach.
- Year
- 2026
- Role
- Frontend engineer
- Status
- Private · client
- Context
- Client
Results
~51
REST operations integrated
Analysis, coaching, payments, social, profiles
A React 19 + Vite single-page app with role-aware routing for players and coaches, lazy-loaded feature modules and around 51 REST operations across swing analysis, coaching, payments, social publishing and profiles.
Swings are captured with the browser camera or uploaded as images and video, then submitted for analysis. Results come back as AI correction sessions with a 3D avatar and skeleton view; the avatar data loads in parallel and downloads as binary .glb.
The networking layer handles Bearer tokens, 30-second timeouts and automatic refresh-token rotation, replaying requests that failed with a 401 while the refresh happened, with retry-loop protection and forced logout when authorization really fails.
It also covers coach approval queues, payment intents and billing history, rewards, and YouTube and TikTok OAuth for publishing swings to social media.