ml-model-serving-playbook

Takes a trained model from artifact to production: a validated FastAPI prediction service, container and Kubernetes packaging, drift monitoring with Prometheus metrics and alerts, MLflow registry promotion, and a CI/CD pipeline with performance gates and rollback. Use when the user says "deploy my model", "put the model behind an API", "set up model monitoring or drift detection", "the prediction endpoint is slow or returning 500s", "wire model training into CI", or "roll out a canary of the new model". Not for training models from scratch, feature engineering, choosing model architectures, or GPU driver and cluster provisioning work.

Version

v1.0.1

Size

28 KB

Updated

2026-10-06

Your client verifies the signature before reading. This explicit manual read is available on every plan. Free automatically loads five chosen skills; paid plans automatically load relevant skills across the library.