MLA-C01 Question 347
Select 4You are a Machine Learning Engineer tasked with deploying a model into production using a CI/CD pipeline on AWS. The pipeline must automate the process of training, testing, and deploying the model. Which of the following are key components you should include to ensure a robust and scalable ML CI/CD workflow on AWS?
- A
AWS CodePipeline for orchestrating the workflow
- B
SageMaker Model Monitor for tracking infrastructure costs
- C
AWS CodeBuild for executing model training scripts
- D
SageMaker Endpoint for hosting and serving the model
- E
Git-based version control for managing code and model artifacts
- F
AWS Lambda for hyperparameter optimization
Show answer and explanation
Correct answers: A, C, D, E
Explanation
A complete CI/CD workflow for machine learning requires tools to orchestrate the pipeline, execute training scripts, track and manage code and model artifacts, and deploy the trained model into production. AWS CodePipeline, AWS CodeBuild, Git-based version control, and SageMaker Endpoints collectively address these requirements. SageMaker Model Monitor and AWS Lambda serve other purposes, such as monitoring model performance and running serverless functions, respectively, but they are not primary components of a CI/CD ML workflow.
- A. Correct.
AWS CodePipeline is essential for orchestrating CI/CD workflows as it integrates various components of the pipeline, such as code, training, and deployment stages.
- B. Incorrect.
SageMaker Model Monitor is used to monitor model performance in production, not to track infrastructure costs. Cost tracking would typically involve AWS Budgets or Cost Explorer.
- C. Correct.
AWS CodeBuild can be used to run custom scripts, such as model training and evaluation tasks, making it a key component of the CI/CD workflow.
- D. Correct.
SageMaker Endpoint is used to deploy and serve the model in production, which is a critical step in the CI/CD pipeline for ML workflows.
- E. Correct.
Git-based version control systems, such as AWS CodeCommit or GitHub, are essential for managing both code and model artifacts to ensure reproducibility and collaboration.
- F. Incorrect.
AWS Lambda is not suitable for hyperparameter optimization. SageMaker's Hyperparameter Tuning Jobs would be the appropriate choice for that task.