🧠 E-cercise Recommender Service

This is the personalized recommendation microservice for E-cercise, powered by FastAPI and Sentence Transformers.

---

🚀 Features

• Equipment option recommendations based on:
  • User goal, weight, height, gender, age
  • Experience level (e.g. Beginner, Athlete)
  • Tag-based preferences
• Embedding-based vector similarity with fallback scoring
• CORS-ready for frontend integration
• Fully containerized with Docker

---

🛠️ Project Structure

.
├── api/                    # FastAPI router
├── services/               # Recommendation logic
├── utils/                  # Vector handling, text utils
├── models/                 # Pydantic schema models
├── data/                   # Precomputed vector cache + equipment options
├── generate_equipment_vector_cache.py  # Precomputes vector cache
├── main.py                 # FastAPI entrypoint
├── config.py               # Loads .env config
├── .env.example            # Example config file
├── requirements.txt        # Python dependencies
├── Dockerfile              # Docker container setup

---

⚙️ Configuration

Copy .env.example → .env and fill in your settings:

cp .env.example .env

---

🐳 Docker Usage

Build

docker build -t recommender .

Run

docker run --env-file .env -p 8000:8000 recommender

---

📦 Regenerate Vector Cache (if equipment changed)

python generate_equipment_vector_cache.py

---

🔌 Example API Request

POST /recommend
Content-Type: application/json

{
  "goal": "weight-loss",
  "experience": "beginner",
  "gender": "female",
  "age": 25,
  "preferences": [
    { "tag": "glutes", "group": "muscle" }
  ]
}

Returns top 10 scored equipment options 🎯

---

🧪 Dev Mode

uvicorn main:app --reload

---

🤝 License

MIT