The iPhone-first fitness command center that unifies training, nutrition, recovery, and body composition into a single privacy-first experience — powered by federated AI.
FitMe replaces your training log, meal tracker, and recovery dashboard with one app that knows what you should do today — without ever seeing your private health data.
🌐 Public showcase: fitme-story.vercel.app — case studies, framework evolution timeline (v1.0 → v7.10 → v8.x build window), and interactive PM-flow diagrams. The story site is the canonical narrative; this repo is the source of truth. For machine-derived current-state counts (framework version, features tracked, instrumented gates) defer to docs/FRAMEWORK-FACTS.md.
Repo name is
FitTracker2for historical reasons. The product brand is FitMe.
Animated demo and device screenshots coming soon. Live story site: fitme-story.vercel.app.
| Home | Training | Nutrition | Stats | Settings |
|---|---|---|---|---|
| coming soon | coming soon | coming soon | coming soon | coming soon |
Design source: FitMe Design System Library on Figma
Training — 87 exercises across a 6-day push/pull/legs split, set-by-set logging with weight/reps/RPE, automatic PR detection with 1RM estimation, floating rest timer with haptics, cardio with heart-rate Zone 2 detection.
Nutrition — Dynamic macros that adapt to training day and program phase, 4-tab meal entry (label OCR with English + Hebrew, manual, templates, barcode via Open Food Facts), supplement streak tracking, hydration with training-day vs rest-day targets.
Recovery & Biometrics — HealthKit (HR, HRV, VO2Max, sleep stages), Xiaomi S400 smart-scale entries (10 metrics), daily readiness score (40% HRV / 30% RHR / 30% Sleep), Recovery Studio with personalized routines.
Stats & Progress — 18 metrics across body, recovery, training, and nutrition; daily / weekly / monthly / 3-month / 6-month views; interactive charts; all-time PRs with Epley 1RM.
AI Intelligence — Three-tier pipeline: local rules → cloud cohort (banded values, k≥50) → on-device Foundation Models (iOS 26+). Confidence-gated, privacy-preserving, graceful offline fallback.
Privacy & Security — Double-layer encryption (AES-256-GCM + ChaCha20-Poly1305 with HMAC-SHA512), Secure Enclave key storage with biometric ACL, zero-knowledge sync (.ftenc blobs only), Apple Sign In + Passkeys (WebAuthn), GDPR-oriented account deletion + data export.
| Layer | Technology |
|---|---|
| UI | SwiftUI, SF Symbols |
| Health | HealthKit |
| Auth | Apple Sign In (Supabase OAuth), Passkeys (WebAuthn), Email/OTP |
| Encryption | AES-256-GCM + ChaCha20-Poly1305 via CryptoKit, HMAC-SHA512 |
| Key Storage | Keychain with biometric ACL, Secure Enclave (P-256) |
| Sync | CloudKit (iCloud Private DB) + Supabase (PostgreSQL + Realtime) |
| AI — Cloud | FastAPI on Railway, JWT + JWKS validation, k≥50 anonymity |
| AI — On-device | Apple Intelligence Foundation Models (iOS 26+) |
| Analytics | Firebase Analytics (GA4) with GDPR consent |
| Design System | 125 semantic tokens, Style Dictionary pipeline, CI drift detection |
iPhone (on-device)
┌─────────────────────────────────────────────────────┐
│ SwiftUI Views │
│ ↕ │
│ EncryptedDataStore ← AES-256-GCM + ChaCha20 │
│ ↕ ↕ │
│ HealthKit Service Keychain / Secure Enclave │
│ ↕ │
│ AI Orchestrator │
│ ├── Local rules (always available) │
│ ├── Cloud cohort (banded values only) ──────────→│── AI Engine (FastAPI)
│ └── Foundation Model (private, on-device) │ k≥50 anonymity
│ │
│ AnalyticsService ← ConsentManager (GDPR) │
│ └── FirebaseAnalyticsAdapter ────────────────────→│── GA4 (Firebase)
└─────────────────────────────────────────────────────┘
↕ encrypted .ftenc blobs only
┌──────────────┐ ┌──────────────┐
│ CloudKit │ │ Supabase │
│ (iCloud) │ │ (PostgreSQL)│
└──────────────┘ └──────────────┘
Zero-knowledge sync: servers store only encrypted .ftenc blobs — no plaintext health data ever leaves the device.
# iOS app — open in Xcode 16+ (iOS 17.0+ deployment target)
open FitTracker.xcodeproj
# Token pipeline (design system)
npm install && npm run tokens:check
# Run a full local verification (tokens + dashboard + iOS targeted tests)
make verify-localFull setup including SSD-relocated build artifacts, web sub-projects, and AI engine: see docs/setup/.
FitMe is a portfolio project built by a Product Manager transitioning into engineering, using Claude Code as a pair-programmer. Every feature was designed, reviewed, and tested by me; AI accelerated implementation. The Co-Authored-By: Claude trailer in commits makes that collaboration visible by design.
The repo runs a custom PM lifecycle (Research → PRD → Tasks → UX → Implement → Test → Review → Merge → Docs) with measurement instrumentation at every phase. Process docs and case studies that show the lifecycle in action live in docs/case-studies/.
| Topic | Link |
|---|---|
| Product PRD + per-feature PRDs | docs/product/ |
| Design system + UX foundations | docs/design-system/ |
| Architecture deep-dive | docs/architecture/ |
| Case studies | docs/case-studies/ |
| Roadmap | docs/master-plan/master-backlog-roadmap.md |
| Changelog | CHANGELOG.md |
MIT © 2026 Regev Barak.