An ML engineer needs to deploy ML models to get inferences from large datasets in an asynchronous manner. The ML engineer also needs to implement scheduled monitoring of the data quality of the…

Question

Options

• ADeploy the models by using scheduled AWS Glue jobs. Use Amazon CloudWatch alarms to
• BDeploy the models by using scheduled AWS Batch jobs. Use AWS CloudTrail to monitor the data
• CDeploy the models by using Amazon Elastic Container Service (Amazon ECS) on AWS Fargate.
• DDeploy the models by using Amazon SageMaker batch transform. Use SageMaker Model Monitor

• A
  16% (3)
• B
  11% (2)
• C
  5% (1)
• D
  68% (13)

Explanation

Option D is correct because SageMaker batch transform is purpose-built for running asynchronous ML inference against large datasets without needing a persistent endpoint, and SageMaker Model Monitor natively supports scheduled data quality checks with built-in CloudWatch alarm integration to alert on drift or quality degradation.

Why the distractors fail:

• A (Glue): AWS Glue is an ETL/data integration service, not an ML inference platform - it has no mechanism to run model predictions.
• B (Batch + CloudTrail): AWS Batch can run compute jobs but lacks native ML inference orchestration; CloudTrail records API calls for auditing, not data quality metrics for models.
• C (ECS/Fargate): While Fargate can host containers, it requires building all monitoring and alerting from scratch - there is no native equivalent to Model Monitor's scheduling or drift detection.

Memory tip: Think "batch in, monitor out" - batch transform handles the input problem (large async datasets) and Model Monitor handles the output problem (watching what happens to data quality over time). Any answer that splits these concerns across non-SageMaker services (Glue, Batch, CloudTrail) is missing the integrated SageMaker solution the question is looking for.

No community discussion yet for this question.