The company requires a manual approval workflow so that only authorized model versions can be deployed to production endpoints. Which solution satisfies this requirement?

SageMaker Model Registry, often integrated with SageMaker Pipelines, allows you to register model versions. Each model version in the registry has an approval status (e.g., "Pending", "Approved", "Rejected"). To implement a manual approval workflow, you would typically register a model version with a "Pending" status. After a human review, an authorized person can then use the AWS SDK (or the AWS Management Console) to programmatically update the model version's status to "Approved." This "Approved" status can then trigger subsequent deployment steps in your CI/CD pipeline. SageMaker Experiments is for tracking training runs, not for managing model deployment approvals. SageMaker ML Lineage Tracking records the lineage of ML artifacts, but doesn't directly manage approval states for deployment. SageMaker Model Monitor evaluates deployed model performance, which is a post-deployment activity, not an approval mechanism for deployment itself.