AWS MLA-C01: AWS Certified Machine Learning Engineer - Associate

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

1. Question 1ML Model Development

   A team wants to start from a pretrained open-source model and fine-tune it on its own data within SageMaker with minimal code. Which capability fits best?

   1. ASageMaker JumpStart
   2. BSageMaker Edge Manager
   3. CAmazon Kinesis Video Streams
   4. DAWS Batch array jobs
   Show answer and explanationHide answer

   Correct answer: A

   JumpStart provides pretrained models and solution templates that can be fine-tuned or deployed with little code. Edge Manager is for devices, Kinesis Video handles streams, and Batch runs general compute jobs.

2. Question 2Data Preparation for Machine Learning

   A team stores training data as thousands of small CSV files in Amazon S3, and training jobs read the data slowly. Which change most improves read throughput?

   1. AConvert the data to larger partitioned columnar files such as Parquet
   2. BIncrease the number of SageMaker instances without changing the files
   3. CMove the files into S3 Glacier Instant Retrieval
   4. DCompress each small file separately with gzip
   Show answer and explanationHide answer

   Correct answer: A

   Many small objects create overhead per read, and partitioned columnar formats like Parquet let jobs read only the needed columns and partitions. Compression alone or more instances do not fix the small-file problem.

3. Question 3Monitoring, Maintenance, and Security

   Which two practices help detect model drift after deployment? (Choose two.)

   1. AIgnore changes in input data
   2. BDisable logging of predictions
   3. CRetrain only once when the model is deployed
   4. DCompare live input distributions against the training baseline
   5. EMonitor prediction quality against ground truth labels when available
   Show answer and explanationHide answer

   Correct answer: D, E

   Comparing live inputs with the training baseline reveals data drift, and tracking quality against labelled outcomes reveals performance drift. Ignoring inputs, disabling logs, and retraining only once remove the signals needed to notice drift.

4. Question 4Deployment and Orchestration of ML Workflows

   A team needs to score a nightly file of ten million records and does not need real-time responses. Which option fits best?

   1. AA WebSocket API in front of the model endpoint
   2. BAn interactive notebook kernel running on a persistent instance
   3. CA batch transform job that reads input from Amazon S3
   4. DA real-time endpoint with autoscaling policies
   Show answer and explanationHide answer

   Correct answer: C

   Batch transform processes large datasets offline from S3 and writes results back, with no endpoint to keep running. A real-time endpoint or WebSocket API is designed for low-latency requests, and notebooks are for interactive work.

5. Question 5ML Model Development

   A tabular classification project needs a strong baseline quickly, with automatic feature processing and model selection. Which SageMaker feature should be used?

   1. ASageMaker Serverless Inference
   2. BSageMaker Ground Truth Plus
   3. CSageMaker Autopilot
   4. DSageMaker Neo
   Show answer and explanationHide answer

   Correct answer: C

   SageMaker Autopilot automatically explores preprocessing and model candidates for tabular data and produces a ranked set of models with notebooks. Neo compiles models for hardware, Serverless Inference hosts models, and Ground Truth Plus manages labelling.