Test

The goal is to automate checks so integration and deployment stay safe. Use tests plus CI (for example GitHub Actions) as gates before changes are merged or published.

In practice:

1. ML project (image-classifier): tests for packaging and inference (load artifact, run predict, expected response shape).
2. MLOps project (image-classifier-pipeline): tests for the served API (health, predict, error handling).
3. CI: run both test suites automatically on push or pull request.

Heavier checks (full evaluate on the test set, benchmark KPIs) can stay manual or scheduled until you promote them to explicit gates.