A transit authority needs Azure Machine Learning training jobs to pull code from a private civic GitHub repository with secure access. What should the MLOps team configure?
Select an answer to reveal the explanation.
Short Explanation
Private civic repos need GitHub integration, since you can't register GitHub as a datastore for mounts. You wire Azure ML to GitHub with secure integration so training pulls stay authenticated and auditable. You'll find managed identity alone won't authenticate to GitHub, and cached environment code won't track your commits.
Full Explanation
Secure GitHub integration with Azure Machine Learning lets transit training jobs authenticate to a private civic repository under governed credentials instead of copying code by hand. Jobs then clone the exact commit the pipeline intends, preserving branch protection, code review, and secret scanning already applied to the municipal repo. Integration avoids baking personal access tokens into every environment variable when federation or managed secrets are configured correctly, and it keeps the training image definition separate from the evolving training scripts.
Enabling the Azure Machine Learning managed identity to pull job code from the private civic GitHub repository fails by authentication type: GitHub does not trust Azure managed identities for git authentication, so jobs still lack a secure GitHub integration to clone the private repository.
Registering the private civic GitHub repository as an Azure Machine Learning datastore for job code mounting fails by service type: datastores connect to Azure storage for data mounting, not to GitHub for source control, and provide no authenticated git clone of the training code.
Caching private civic GitHub repository code in the Azure Machine Learning environment so jobs reuse an installed copy fails on lifecycle: baked-in code goes stale, forces a rebuild for every commit, and loses the authenticated, commit-pinned clone that secure GitHub integration provides.
Exam caveat: AI-300 stresses secure source-control integration for private repositories used by Azure ML jobs; any answer that relies on managed identity alone for GitHub, registers GitHub as a datastore, or caches code in the environment instead of integrating with the private GitHub repo is wrong when the stem specifies a private civic GitHub repository. Operational check: trigger a training job that references the private repo, confirm the job log shows a successful authenticated clone of the expected branch or commit, and verify that repository access policies still deny anonymous clone from the public internet.
Prefer to study offline? Printable PDF study book + test engine ($7.50, unofficial). These questions stay free here.