A county wants ephemeral Azure Machine Learning experiment workspaces removed automatically after short-lived grant pilots to control cost. Which approach best supports that lifecycle?

Automating create and teardown of temporary civic ML environments through the same GitHub Actions IaC pipeline gives the county an ephemeral lifecycle for grant pilots. Provisioning and destroy share one definition: when the pilot ends, teardown removes the Azure Machine Learning workspace and related spend instead of relying on memory. Cost control and reduced orphaned resources follow directly from treating temporary environments as code with an explicit end.

Automating start and stop schedules for pilot ML compute resources through the same GitHub Actions IaC pipeline only idles the compute; the workspace itself remains provisioned and billable, so it does not remove the ephemeral environment when the pilot ends. Automating cost alerts and dashboard reporting for pilot ML spending through the same GitHub Actions IaC pipeline merely reports on spend after the fact; it provides no mechanism to tear down the workspace automatically when the grant pilot concludes. Automating export and versioning of pilot workspace artifacts through the same GitHub Actions IaC pipeline preserves the experiment outputs rather than removing the workspace itself, so it provides no mechanism to tear down the ephemeral environment when the pilot ends.

Exam caveat: When stems emphasize ephemeral environments and cost control, the correct pattern is automated provision/teardown via IaC pipelines—not start and stop schedules, cost alerts and dashboards alone, or exporting and versioning artifacts. Operational check: create a pilot environment with the workflow, confirm resources and tags, run teardown, verify the resource group and workspace are gone, and check Cost Management that charges stop accruing for that pilot tag.