Question

Your organization has multiple Azure Machine Learning workspaces used by different data science teams. You have built a validated data-preprocessing component that all teams should reuse in their pipelines. You want to publish it once so that any workspace in the organization can reference the exact same versioned component without copying files between workspaces. Which approach should you use?

Accepted Answer

Create an Azure Machine Learning registry and publish the component to it, then reference it from each workspace's pipeline jobs

Answer

Register the component in one workspace and grant the other workspaces' managed identities Reader access to that workspace

Answer

Store the component YAML and scripts in a shared blob datastore and have each team register it locally in their own workspace

Answer

Export the component as a Docker image and push it to Azure Container Registry for all teams to pull