MLA-C01 Question 355
Select 3A data science team is using AWS CodeCommit as their code repository and AWS CodePipeline to automate the deployment of machine learning models. They want to ensure that every code commit triggers a pipeline execution that builds the model, runs tests, and deploys the model to an endpoint in Amazon SageMaker. Which of the following steps are necessary to set up this workflow?
- A
Create a CodePipeline pipeline and configure an AWS CodeCommit repository as the source.
- B
Use an Amazon SageMaker notebook instance to manually trigger the CodePipeline execution.
- C
Add a build stage in CodePipeline that uses AWS CodeBuild to build and package the model.
- D
Configure a deploy stage in CodePipeline to deploy the model to an Amazon SageMaker endpoint.
- E
Enable pipeline notifications using Amazon SNS to monitor code commit events.
Show answer and explanation
Correct answers: A, C, D
Explanation
To automate the deployment of machine learning models using AWS CodePipeline, you must configure a source stage (e.g., CodeCommit), a build stage (e.g., CodeBuild), and a deploy stage (e.g., SageMaker endpoint deployment). This setup ensures that every repository update triggers the pipeline for end-to-end automation. Optional features like Amazon SNS for notifications can enhance monitoring but are not required to complete the pipeline workflow.
- A. Correct.
Correct. Configuring AWS CodeCommit as the source ensures that every commit to the repository triggers the pipeline execution.
- B. Incorrect.
Incorrect. Manually triggering the execution defeats the purpose of automation. CodePipeline is designed to handle this step automatically.
- C. Correct.
Correct. CodeBuild is necessary to build and package the machine learning model during the build stage of the pipeline.
- D. Correct.
Correct. A deploy stage in CodePipeline is required to deploy the trained machine learning model to an Amazon SageMaker endpoint.
- E. Incorrect.
Incorrect. While Amazon SNS can provide notifications, it is not required for setting up the pipeline workflow. Notifications are optional for monitoring purposes.