🧾 Project Prompt (Assignment)

Crea una API en Python que utilice un modelo entrenado para hacer predicciones. La API recibe datos y devuelve resultados.

Stakeholders want your classifier available via an API so other apps can use it.

Expose your trained model as a versioned API with a /predict endpoint.

Requirements

• Load the model artifact on startup
• Validate the request schema (Pydantic)
• Return prediction + probability/confidence
• Add versioning: /v1/predict
• Add /health endpoint
• Add request logging (request id, latency)
• Add tests for predict (valid + invalid payloads)

🛠 Tools (Tech Stack)

• Python + FastAPI + Uvicorn
• scikit-learn (or your chosen model)
• Pydantic
• Pytest

💡 Implementation Hints

• Keep preprocessing identical to training (Pipeline)
• Don’t reload the model per request (load once)