Overview
Designed and built end-to-end a native iOS fitness tracker combining computer vision, on-device AI, and depth sensing. The app uses Vision framework pose estimation and a self-trained CoreML classifier to count reps in real time through the device camera. An AI coach powered by Apple Foundation Models runs entirely on-device to analyze session and routine progress, providing personalized insights without sending data to external servers. LiDAR sensor integration enables automatic body measurement scanning for precise physical tracking. Additional features include workout session logging, saved routines with weekly scheduling, personal records, progress charts with Swift Charts, and automatic iCloud Drive backups every 6 hours. Architected with Clean Architecture (MVVM + Repository Pattern + Use Cases + DI container), SwiftData persistence, and built entirely with native Apple frameworks — zero third-party dependencies.