Machine Learning Engineer Associate · 24% of the exam

ML Solution Monitoring, Maintenance, and Security: free practice questions

5 sample questions from our 14-question bank for this domain — answers and explanations included. These are the same scenario-based style as the real AWS exam.

1. To monitor an endpoint's operational metrics (latency, invocation counts, errors, CPU/memory) and set alarms, which AWS service should be used?

  • A. Amazon CloudWatch✓ Correct
  • B. AWS Artifact
  • C. Amazon Macie
  • D. AWS Direct Connect
Explanation

Amazon CloudWatch collects operational metrics and logs from SageMaker endpoints and supports alarms and dashboards. Artifact (B) is compliance docs, Macie (C) finds sensitive data, and Direct Connect (D) is networking — none monitors operational endpoint metrics.

2. For a security audit, a team needs a record of every API call made to SageMaker and who made it. Which service provides this audit trail?

  • A. AWS CloudTrail✓ Correct
  • B. Amazon CloudWatch metrics only
  • C. Amazon SNS
  • D. Amazon S3 Transfer Acceleration
Explanation

AWS CloudTrail logs API activity across the account — who called which API, when, and from where — the audit trail for governance. CloudWatch metrics (B) show performance, not a full API audit; SNS (C) is messaging; and S3 Transfer Acceleration (D) speeds uploads — neither audits API calls.

3. An endpoint shows rising latency and error rates under load. Using CloudWatch, what is an appropriate FIRST response to maintain service quality?

  • A. Set CloudWatch alarms and enable/adjust auto scaling to add capacity, then investigate the root cause✓ Correct
  • B. Ignore the alarms because latency doesn't matter
  • C. Immediately delete the model from the registry
  • D. Turn off CloudWatch to stop the alarms
Explanation

Alarms plus auto scaling add capacity to protect service quality while you investigate the underlying cause. Ignoring the problem (B) harms users, deleting the model (C) is destructive and premature, and turning off monitoring (D) hides the issue rather than fixing it.

4. Which combination BEST represents securing an end-to-end ML system on AWS?

  • A. Least-privilege IAM roles, KMS encryption at rest and TLS in transit, VPC isolation, and CloudTrail/CloudWatch monitoring✓ Correct
  • B. Public buckets, shared credentials, and no logging
  • C. Only encrypting the model file and nothing else
  • D. Relying on the algorithm to secure the infrastructure
Explanation

End-to-end ML security is layered: least-privilege IAM, encryption at rest (KMS) and in transit (TLS), network isolation (VPC), and monitoring/audit (CloudWatch, CloudTrail). Option B lists anti-practices, option C is insufficient, and option D wrongly assumes the model secures the surrounding system.

5. Which SageMaker capability provides ongoing bias monitoring and feature-attribution (explainability) drift detection for models in production?

  • A. Amazon SageMaker Clarify (integrated with Model Monitor)✓ Correct
  • B. Amazon SageMaker Neo
  • C. AWS Trusted Advisor
  • D. Amazon QuickSight
Explanation

SageMaker Clarify supports bias detection and explainability, and integrates with Model Monitor to watch for bias drift and feature-attribution drift in production. Neo (B) optimizes models for hardware, Trusted Advisor (C) gives account best-practice checks, and QuickSight (D) is BI — none does production bias/explainability monitoring.

9 more questions in this domain

Practice the full bank with instant grading, flashcards, and a timed mock exam.

Start practicing free