Summary
In this module, you took a trained model from registration through automated deployment to production monitoring in Azure Machine Learning.
What you learned
- You register and version MLflow models, then archive versions that no longer need to appear in default lists.
- You explore the MLOps stages that connect model development to deployment, automation, and monitoring.
- You use protected GitHub environments to scope deployment access and require approval before production promotion.
- You deploy a model to a managed online endpoint and use a blue-green rollout to test a new version before shifting production traffic to it.
- You troubleshoot deployment and scoring failures by checking deployment details and container logs.
- You keep the previous deployment available so you can roll back by redirecting traffic without changing the endpoint URL.
- You automate registration, deployment, and testing with GitHub Actions by using OIDC federated credentials.
- You use operational and model monitoring signals to decide when to roll back, investigate, or retrain.