An ML engineer is developing a classification model. The ML engineer needs to use custom libraries in processing jobs, training jobs, and pipelines in Amazon SageMaker. Which solution will provide…

Question

Options

• AManually install the libraries in the SageMaker containers.
• BBuild a custom Docker container that includes the required libraries. Host the container in
• CCreate a SageMaker notebook instance to host the jobs. Create an AWS Lambda function to
• DRun code for the libraries externally on Amazon EC2 instances. Store the results in Amazon S3.

• A
  4% (1)
• B
  78% (18)
• C
  13% (3)
• D
  4% (1)

Explanation

Building a custom Docker container with the required libraries pre-installed is the lowest-effort solution because SageMaker natively supports custom containers across processing jobs, training jobs, and pipelines - you build once, register it in Amazon ECR, and reference it everywhere without any per-job setup.

Why the distractors fail:

• A - Manually installing libraries each time a container starts requires scripting the installation into every job definition and reruns on every execution, adding repetitive overhead and fragility.
• C - Hosting jobs on a notebook instance misuses that service (notebooks are for interactive development, not production pipelines), and adding a Lambda function introduces unnecessary architectural complexity.
• D - Offloading library execution to EC2 defeats the purpose of SageMaker's managed infrastructure and requires you to build and maintain a separate compute layer plus S3 handoff logic.

Memory tip: Think "build once, run anywhere" - a Docker container is the standard, portable packaging unit. Whenever an AWS exam asks about custom dependencies with least effort in a managed service context, a pre-built container registered in ECR is almost always the answer because it integrates cleanly without fighting the service's design.

No community discussion yet for this question.