AWS MLA-C02: AWS Certified Machine Learning Engineer - Associate (updated, beta)

Last checked October 22, 2026 against the official AWS exam page.

1. Question 1ML Model Development

   A team wants SageMaker to automatically try multiple algorithms and preprocessing steps on a tabular dataset and produce a ranked leaderboard with explanations. Which feature should they use?

   1. ASageMaker Edge Manager
   2. BSageMaker Pipelines
   3. CSageMaker Serverless Inference
   4. DSageMaker Autopilot
   Show answer and explanationHide answer

   Correct answer: D

   SageMaker Autopilot automatically explores candidate pipelines, ranks them on a leaderboard, and can generate notebooks that explain the process. Edge Manager manages models on devices, Pipelines orchestrates workflows, and Serverless Inference hosts models.

2. Question 2Data Preparation for ML

   A data scientist must clean and transform a 500 GB CSV dataset stored in Amazon S3, removing duplicates and imputing missing values through a visual interface, without writing code. Which SageMaker capability should be used?

   1. ASageMaker Ground Truth
   2. BSageMaker Data Wrangler
   3. CSageMaker Feature Store
   4. DSageMaker Model Monitor
   Show answer and explanationHide answer

   Correct answer: B

   SageMaker Data Wrangler provides a visual, no-code interface for data preparation, including deduplication and missing-value handling, and it can export the flow to a processing job. Model Monitor watches deployed models, Ground Truth manages labeling, and Feature Store stores features.

3. Question 3Deployment and Orchestration of ML Workflows

   A model must serve inference requests with payloads up to one gigabyte and processing times of several minutes, with results delivered when ready and no need to keep a persistent connection open. Which SageMaker inference option is most suitable?

   1. AReal-time endpoint with GPU
   2. BServerless endpoint
   3. CBatch transform job
   4. DAsynchronous inference
   Show answer and explanationHide answer

   Correct answer: D

   Asynchronous inference queues requests, supports large payloads and long processing times, and scales to zero when idle. Real-time endpoints need a persistent low-latency response, serverless inference has tighter payload and duration limits, and batch transform processes an entire dataset offline.

4. Question 4ML Solution Monitoring, Maintenance and Security

   A deployed model's input feature distributions have shifted from the training baseline, and the team wants to be alerted automatically. Which SageMaker capability should be configured?

   1. ASageMaker Model Monitor with a data quality baseline and scheduled monitoring jobs
   2. BSageMaker Ground Truth active learning that selects uncertain samples for human annotation
   3. CSageMaker Autopilot leaderboard that ranks candidate models by validation objective metric
   4. DSageMaker Neo compilation logs written to CloudWatch for each compiled model artifact version
   Show answer and explanationHide answer

   Correct answer: A

   Model Monitor compares live data with a baseline created from training data and raises alerts on violations such as drift, through scheduled monitoring jobs. The other options do not monitor production data quality.

5. Question 5ML Model Development

   A data scientist wants to tune the hyperparameters of a training job efficiently across a large search space by using prior results to choose the next candidates. Which SageMaker feature is designed for this?

   1. AIncreasing the instance size until accuracy improves
   2. BAutomatic model tuning with the Bayesian search strategy
   3. CRandom sampling of the training data only
   4. DA manual grid of fixed values run sequentially in a notebook
   Show answer and explanationHide answer

   Correct answer: B

   Automatic model tuning supports Bayesian optimization, which uses results from earlier trials to choose promising hyperparameters. A sequential manual grid is inefficient, sampling data does not tune hyperparameters, and instance size does not search the hyperparameter space.