Rockie

AI-powered rock climbing analysis app. Record a climb, get your efficiency score, pose breakdown, and actionable feedback — all on your phone.

---

What it does

• Upload a climb video from your library or record directly
• CV pipeline extracts pose data frame-by-frame using MediaPipe
• Efficiency score (0–100) based on hip drops, barn doors, foot swaps, arm shake
• Annotated clips — full, crux, and best sequence highlights
• AI feedback — specific, actionable coaching based on your movement patterns
• Progress dashboard — track scores over time with trend chart

---

Stack

Layer	Tech
iOS App	Expo (React Native) + Expo Router + Clerk + Supabase
CV Worker	FastAPI + MediaPipe Tasks + FFmpeg
Database	Supabase (PostgreSQL + Storage)
Deployment	Railway (CV worker)
Auth	Clerk

---

Repo structure

rockie/
├── rockie-app/          # Expo iOS app
│   ├── app/             # Expo Router screens
│   ├── lib/             # Supabase + API clients
│   └── components/      # Shared components
├── cv-worker/           # FastAPI CV pipeline
│   ├── main.py          # API entrypoint
│   ├── services/        # pose_extractor, scorer, clip_annotator, etc.
│   └── Dockerfile
└── supabase/
    └── migrations/      # DB schema

---

Running locally

CV Worker

cd cv-worker
python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt
cp .env.example .env   # add SUPABASE_URL + SUPABASE_SERVICE_KEY
python3 -m uvicorn main:app --host 0.0.0.0 --port 8000

iOS App

cd rockie-app
npm install
cp .env.example .env.local   # add Supabase + Clerk keys
npx expo start

---

Environment variables

cv-worker/.env

SUPABASE_URL=
SUPABASE_SERVICE_KEY=

rockie-app/.env.local

EXPO_PUBLIC_SUPABASE_URL=
EXPO_PUBLIC_SUPABASE_ANON_KEY=
EXPO_PUBLIC_CLERK_PUBLISHABLE_KEY=
EXPO_PUBLIC_CV_WORKER_URL=