MLA-C01 Question 412
Select 3You are a Machine Learning Engineer working for a retail company that uses an Amazon SageMaker model to predict demand for products. You need to ensure compliance by tracking and auditing all activities related to model re-training, including who initiated the re-training and when it occurred. Additionally, you want to automate the re-training process if a specific event, such as a drop in model accuracy, is detected. How can AWS CloudTrail and other AWS services be used to accomplish this?
- A
Enable AWS CloudTrail to log API calls for SageMaker and analyze the logs to track re-training activities.
- B
Set up an Amazon CloudWatch Event rule to monitor CloudTrail logs for specific SageMaker events and trigger re-training workflows.
- C
Use AWS CloudTrail Insights to automatically trigger the re-training process whenever unusual API activity is detected for SageMaker.
- D
Store CloudTrail logs in Amazon S3 and use Amazon Athena to query logs for compliance and auditing purposes.
- E
Configure AWS CloudTrail to directly invoke an Amazon SageMaker re-training job whenever a re-training event is logged.
Show answer and explanation
Correct answers: A, B, D
Explanation
To log, monitor, and invoke re-training activities, AWS CloudTrail can be used to log SageMaker API calls, enabling you to track who initiated re-training and when it occurred. For automation, CloudWatch Event rules can monitor CloudTrail logs for specific events and trigger workflows like re-training. Storing these logs in S3 and querying them with Athena helps with compliance and auditing. AWS CloudTrail itself cannot directly trigger SageMaker jobs, and CloudTrail Insights is not relevant for this use case.
- A. Correct.
Enabling AWS CloudTrail to log API calls for SageMaker is necessary to track activities such as who initiated re-training and when it occurred. This is a key step for compliance and auditing.
- B. Correct.
Amazon CloudWatch Event rules can monitor CloudTrail logs for specific SageMaker events and trigger workflows, such as invoking a Lambda function to start a re-training process. This helps automate the re-training process.
- C. Incorrect.
AWS CloudTrail Insights is used for detecting unusual API activity, but it is not designed to trigger re-training processes specifically and is unrelated to this scenario.
- D. Correct.
Storing CloudTrail logs in Amazon S3 and querying them with Amazon Athena provides an efficient way to perform compliance checks and auditing on re-training activities.
- E. Incorrect.
AWS CloudTrail does not have the capability to directly invoke SageMaker jobs. It can log events, but you need an intermediary service like AWS Lambda for execution.