← All projects
Machine learning & AI · Python
M

ML Model Serving API

Build a POST /predict payload, see validation reject bad input at the edge, and switch model versions with a rollback — the production half of scikit-learn serving.

01 Model versions

Each bundle carries its own features, metrics and classes. Activate one to serve it — no restart, the way POST /model/activate/{version} works.

Active model metadata
FieldValue

02 Request — POST /predict

Pydantic rejects empty, non-numeric, NaN and oversized inputs; the wrong number of features returns 422 with the count it expected.

03 Response

—

{ }

04 Ops endpoints