A company is building a model with Amazon SageMaker and must track model bias and show results on a dashboard. After creating a bias monitoring job, how should the engineer capture the bias metrics for dashboarding?

The correct answer is to collect Amazon CloudWatch metrics emitted by SageMaker Clarify. SageMaker Clarify, which is used for bias detection and explainability, automatically publishes its bias metrics to Amazon CloudWatch. These metrics can then be easily visualized and monitored on a CloudWatch dashboard. Collecting AWS CloudTrail events from SageMaker Clarify would show API call history, not the bias metrics themselves. SageMaker Model Monitor is used for monitoring model quality, data quality, and concept drift in production, not for capturing bias metrics from SageMaker Clarify. While EventBridge and SNS can be used for notifications and event routing, they are not the primary mechanism for directly collecting and dashboarding the bias metrics generated by SageMaker Clarify; CloudWatch is the direct integration for these metrics.