Machine Learning Engineer Associate Practice Questions
Prepare for MLA-C01 with more than an answer.
- Exam fee
- $150 USD
- Time limit
- 130 minutes
- Passing score
- 720
- Level
- Associate
- Valid for
- 3 years
Domains covered on the exam 4
- Data Preparation for Machine Learning (ML)28%
- ML Model Development26%
- Deployment and Orchestration of ML Workflows22%
- ML Solution Monitoring, Maintenance, and Security24%
- 1
A startup is deploying a Large Language Model (LLM) for a chatbot application. The traffic pattern is highly unpredictable: there are long periods of inactivity followed by sudden bursts of user requests. The model size is 10 GB. Cost optimization is a primary concern, and cold-start latency of a few seconds is acceptable for the first request in a burst. Which SageMaker Inference option is MOST cost-effective?
Show answer details
Correct answer: C
SageMaker Serverless Inference is designed for workloads with intermittent or unpredictable traffic. It automatically provisions compute capacity based on the volume of inference requests and scales down to zero when idle, charging only for the compute time used. The tolerance for cold starts aligns with Serverless Inference characteristics.
- 2
A healthcare organization is training a model to predict patient readmissions. They must ensure the model does not exhibit bias against specific demographic groups (e.g., Age, Gender). The dataset is stored in S3. Which tool should the ML Engineer use to generate a pre-training bias report to analyze class imbalance and difference in proportions?
Show answer details
Correct answer: B
SageMaker Clarify provides bias detection capabilities across the ML lifecycle. It can run processing jobs on data in S3 to generate pre-training bias reports, calculating metrics like Class Imbalance (CI) and Difference in Proportions of Labels (DPL).
- 3
True or False: When using Amazon SageMaker Managed Spot Training, the training job can be interrupted if AWS needs to reclaim the capacity, but SageMaker can automatically resume the job from the last saved checkpoint if configured correctly.
Show answer details
Correct answer: A
This is True. Managed Spot Training uses Spot Instances to reduce cost. AWS can interrupt these instances with a 2-minute warning. SageMaker handles the interruption and can restart the job. To avoid losing progress, the training script must implement checkpointing (saving model state to S3) so it can resume from the last checkpoint rather than starting over.
- 4
A Machine Learning Engineer needs to deploy a model to an edge device that has limited compute and memory resources. The model was trained using PyTorch. Which AWS service should be used to compile and optimize the model specifically for the target edge hardware architecture?
Show answer details
Correct answer: B
Amazon SageMaker Neo enables developers to optimize machine learning models for inference on cloud instances and edge devices. It compiles models into an executable that is optimized for the specific hardware (CPU, GPU, or AI accelerator) of the target device.
- 5
An e-commerce company uses a SageMaker endpoint to serve product recommendations. The team wants to ensure that if the data distribution of the incoming requests changes significantly compared to the training data (data drift), they are alerted. Which combination of services should be used to achieve this? (Select TWO)
Show answer details
Correct answer: A, B
SageMaker Model Monitor captures input/output data from the endpoint and compares it against a baseline to detect drift.
Model Monitor publishes metrics to CloudWatch. CloudWatch Alarms can be configured to watch these metrics and send alerts (e.g., via SNS) when drift thresholds are breached.
- 6
A Machine Learning Engineer needs to define an IAM role for a SageMaker Notebook Instance. The data scientists using the notebook need access to specific S3 buckets for training data and model artifacts. Following the principle of least privilege, which policy configuration is BEST?
Show answer details
Correct answer: B
This adheres to the principle of least privilege by restricting actions to only what is necessary (List/Get/Put) and scoping the resource to specific bucket ARNs, rather than granting full access to all buckets.
- 7
Case Study -A company is building a web-based AI application by using Amazon SageMaker. The application will provide the following capabilities and features: ML experimentation, training, a central model registry, model deployment, and model monitoring.The application must ensure secure and isolated use of training data during the ML lifecycle. The training data is stored in Amazon S3.The company needs to use the central model registry to manage different versions of models in the application.Which action will meet this requirement with the LEAST operational overhead?
Show answer details
Correct answer: C
- 8
Case Study -A company is building a web-based AI application by using Amazon SageMaker. The application will provide the following capabilities and features: ML experimentation, training, a central model registry, model deployment, and model monitoring.The application must ensure secure and isolated use of training data during the ML lifecycle. The training data is stored in Amazon S3.The company is experimenting with consecutive training jobs.How can the company MINIMIZE infrastructure startup times for these jobs?
Show answer details
Correct answer: B
