PushupTax: Gamifying Fitness with Computer Vision
I built PushupTax to turn scrolling into a "tax" paid in pushups.
Technical Highlights
- Computer Vision: Used VisionKit (
VNDetectHumanBodyPoseRequest) to track body joints. I engineered the logic to detect a full pushup cycle—plank, descent, and ascent—ensuring counts are accurate and reliable regardless of camera angle. - Backend & Security: Powered by Supabase and PostgreSQL. I implemented Row Level Security (RLS) to protect user data and wrote PL/pgSQL database functions and triggers to handle streak calculations and data automation server-side.
- Modern SwiftUI: Built with a clean MV architecture as opposed to MVVM. I leveraged the @Observable macro to simplify state management and keep the codebase lean and performant.
- System Integration: Integrated AppIntents so users can log sessions via Siri or Shortcuts.
- Production Essentials: Handled secure token storage via Keychain, user authentication through Supabase, and implemented StoreKit 2 for in-app purchases.
This was the first app I decided and developed from scratch, making all architectural, backend, and frontend decisions (with help from a UI/UX designer).
Screenshots