AWS Machine Learning Engineer Associate (MLA-C01) · 24% practice allocation

AWS MLA-C01: ML Solution Monitoring, Maintenance, and Security Practice Questions

5 free sample questions from the ML Solution Monitoring, Maintenance, and Security domain, with answers and explanations. The full MLA-C01 bank has 59 questions across 4 domains. Questions are scenario-based and written from the official AWS exam guide.

1.

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 S3 Transfer Acceleration
  • C. Amazon SNS
  • D. Amazon CloudWatch metrics only
Explanation

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

2.

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

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

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

3.

A team wants to control ML costs. Which practices directly help? (Select 2)

  • A. Duplicate every endpoint across all Regions unnecessarily
  • B. Never monitor spend
  • C. Use auto scaling and right-sized instances; shut down idle resources✓ Correct
  • D. Run all endpoints on the largest GPU instances 24/7 regardless of load
  • E. Choose serverless or batch inference for intermittent or offline workloads✓ Correct
Explanation

Right-sizing with auto scaling and shutting down idle resources (C), and matching inference type to workload — serverless for intermittent, batch for offline (E) — control cost. Always-on max GPU instances (D), ignoring spend (B), and needless multi-Region duplication (A) inflate cost.

4.

A deployed model's accuracy has quietly degraded because the real-world input distribution has shifted since training. What is this phenomenon, and which SageMaker capability detects it?

  • A. Underfitting — detected by Amazon CloudFront
  • B. Data drift — detected by Amazon SageMaker Model Monitor✓ Correct
  • C. Class imbalance — detected by Amazon Route 53
  • D. Overfitting — detected by AWS Artifact
Explanation

When production input distributions drift from training data, model quality degrades — data drift. SageMaker Model Monitor detects drift in data and model quality on live endpoints. The other pairings misname the phenomenon and match it to unrelated services (compliance docs, DNS, CDN).

5.

What is the difference between DATA drift and CONCEPT (model quality) drift?

  • A. They are the same thing
  • B. Data drift is a change in the input distribution; concept drift is a change in the relationship between inputs and the target✓ Correct
  • C. Concept drift means the endpoint was deleted
  • D. Data drift only affects images; concept drift only affects text
Explanation

Data drift is when the input feature distribution shifts; concept/model-quality drift is when the underlying input-to-target relationship changes, degrading predictions even if inputs look similar. Options A, C, and D are incorrect definitions.

Test yourself across every MLA-C01 domain

The free MLA-C01 practice test mixes about 20 questions weighted like the exam. The full 59-question bank adds instant grading, flashcards and a timed mock exam.

Take the free MLA-C01 practice testPractice the full bank

Study this domain

ML Solution Monitoring, Maintenance, and Security study guide (MLA-C01) →

Other MLA-C01 domains

← All AWS Machine Learning Engineer Associate study resources