Skip to main content

Fitness Tracker

Build a workout + activity tracker with a Wear OS companion. This project owns sensors, Health Connect, background services, KMP sharing, and cross-device sync.

The user journey​

Phone                                    Watch
─────                                    ─────
Dashboard (today's stats)                Workout list
Start workout ──────── BLE sync ────────►Live session (heart rate, splits)
End workout ◄───────── sync result ─────End workout
History (trends, charts)

Features (by milestone)​

M1 — Phone skeleton​

  • Bottom nav: Today, Workouts, Trends, Profile
  • Permissions: Activity Recognition, Body Sensors, Health Connect
  • DataStore Proto schema for user preferences + workout settings
  • Compose charts (Vico or custom Canvas) for dashboards

M2 — Sensors & live workout​

  • Sensor fusion: accelerometer + step counter + heart rate
  • Foreground service with notification showing live workout stats
  • Rolling buffer of recent samples for graph rendering
  • Pause/resume/stop with persisted state (survives process death)
class WorkoutService : LifecycleService(), SensorEventListener {
    private lateinit var sensorManager: SensorManager
    private val buffer = RingBuffer<SensorSample>(capacity = 1000)

    override fun onSensorChanged(event: SensorEvent) {
        buffer.add(SensorSample(event.sensor.type, event.values, event.timestamp))
    }

    override fun onStartCommand(intent: Intent?, flags: Int, startId: Int): Int {
        startForeground(NOTIF_ID, buildNotification(), FOREGROUND_SERVICE_TYPE_HEALTH)
        registerSensors()
        lifecycleScope.launch {
            while (isActive) {
                delay(1000)
                val snapshot = buffer.snapshot()
                statsStore.update(snapshot)
            }
        }
        return START_STICKY
    }
}

M3 — Health Connect integration​

  • Read existing data: steps, distance, active calories, heart rate
  • Write workouts: ExerciseSessionRecord with start/end, type, duration, route
  • Permission rationale UI with explicit scopes
  • Sync Health Connect → your cloud backend (debounced)
class HealthSync @Inject constructor(private val client: HealthConnectClient) {
    suspend fun writeWorkout(workout: Workout) {
        val session = ExerciseSessionRecord(
            startTime = workout.startTime, startZoneOffset = ZoneOffset.UTC,
            endTime = workout.endTime, endZoneOffset = ZoneOffset.UTC,
            exerciseType = ExerciseSessionRecord.EXERCISE_TYPE_RUNNING,
            title = workout.title,
            metadata = Metadata.manualEntry()
        )
        val heartRate = HeartRateRecord(
            startTime = workout.startTime, startZoneOffset = ZoneOffset.UTC,
            endTime = workout.endTime, endZoneOffset = ZoneOffset.UTC,
            samples = workout.heartRateSamples.map {
                HeartRateRecord.Sample(time = it.time, beatsPerMinute = it.bpm.toLong())
            }
        )
        client.insertRecords(listOf(session, heartRate))
    }
}

M4 — Wear OS companion​

  • Separate :wear module with Wear Compose UI
  • Tile (ambient stats) + Complication (step count on the watchface)
  • Data Layer API for phone ↔ watch sync
  • Wear-specific foreground service for workouts started on the watch
// Wear: start workout, send to phone
class WatchWorkoutViewModel @Inject constructor(
    private val nodeClient: NodeClient,
    private val messageClient: MessageClient
) : ViewModel() {
    fun sendWorkoutStart() = viewModelScope.launch {
        val nodes = nodeClient.connectedNodes.await()
        val payload = ProtoBuf.encodeToByteArray(WorkoutStart.serializer(), WorkoutStart(now()))
        nodes.forEach { node -> messageClient.sendMessage(node.id, "/workout/start", payload) }
    }
}

M5 — Trends, goals, gamification​

  • Weekly/monthly rollups in a background WorkManager job
  • Goals: steps/day, workouts/week, active minutes
  • Achievements with on-device logic (no server round-trip)
  • Compose animations celebrating streak milestones

M6 — Production​

  • Battery profiling (Battery Historian)
  • Doze + App Standby + sensor batching for efficiency
  • Baseline profile for dashboard and live workout screens
  • Play Data Safety: Health data, body sensor declarations with user controls
  • HIPAA-ish hardening: encryption at rest (SQLCipher) + transit (cert pinning)

Architecture — where the watch sits​

┌─────────────┐  Data Layer API   ┌─────────────┐
│  Phone app  │ ◄────────────────► │  Watch app  │
│  (Compose)  │                    │ (Wear UI)   │
└──────┬──────┘                    └──────┬──────┘
       │                                  │
       ▼                                  ▼
┌─────────────────────────────────────────────────┐
│  :shared (KMP commonMain)                       │
│    Domain models, formatters, formulas          │
└──────────────────────────┬──────────────────────┘
                           │
            ┌──────────────┴──────────────┐
            ▼                             ▼
     ┌─────────────┐              ┌──────────────┐
     │ Health Connect│              │  Cloud API   │
     └─────────────┘              └──────────────┘

Stretch goals​

🏃

GPS route tracking

Phone-only runs with route polyline, split analysis, elevation graph.

💧

Hydration reminders

WorkManager-driven reminders with rich notifications and one-tap log.

🧘

Guided breathing

Compose animation synced to inhale/exhale cycles. Wear tile shortcut.

🔊

Voice coach

TTS split announcements during runs. Audio ducks music automatically.

🎨

Ambient mode

Wear app supports always-on display with low-power updates every 60s.

Testing strategy​

  • SensorManager fakes for unit tests of sample processing
  • Health Connect test doubles for record assertions
  • Robolectric for WorkManager periodic jobs
  • Espresso-Wear for watch app UI tests on emulator
  • MockDataLayer for phone↔watch message routing

Battery SLOs​

  • Live workout: < 10% battery drain per hour on Pixel 7
  • Idle background: < 1% battery drain per 24h
  • Wear companion: < 15% drain per 8h day with 1h workout

Privacy (critical for health apps)​

  • All health data encrypted on device (SQLCipher wrapping Room)
  • Zero third-party SDKs receive health data (no Firebase Analytics on health events)
  • Export + delete flows for GDPR/CCPA compliance
  • BAA required with any cloud provider if marketing as HIPAA-compliant

Next​

Continue to Project 5 — News Aggregator for Paging 3, offline-first content, and Glance widgets.