MLA-C01 Question 320
Select 3You are a Machine Learning Engineer working on deploying a model to production using Amazon SageMaker. To ensure a robust deployment process, you need to set up a CI/CD pipeline that trains the model, validates the results, and deploys it. Which of the following steps should be included in your CI/CD pipeline using AWS services?
- A
Use AWS CodePipeline to orchestrate the CI/CD workflow.
- B
Use Amazon SageMaker Model Monitor to detect changes in input data distribution.
- C
Use AWS CodeBuild to package and containerize the ML model code.
- D
Use AWS Lambda to deploy the model directly to production.
- E
Use Amazon SageMaker Training Jobs to retrain the model as part of the pipeline.
Show answer and explanation
Correct answers: A, C, E
Explanation
In a CI/CD pipeline for machine learning, AWS CodePipeline is used to orchestrate the entire workflow, AWS CodeBuild is used for packaging and building the code, and Amazon SageMaker Training Jobs are used to retrain the model as part of the pipeline. While Amazon SageMaker Model Monitor is important for production monitoring, it is not directly involved in the CI/CD pipeline setup. Similarly, AWS Lambda is not the appropriate service for deploying models in a CI/CD context; Amazon SageMaker Endpoints would typically handle deployment tasks.
- A. Correct.
Correct: AWS CodePipeline is a fully managed CI/CD service that can orchestrate workflows, making it a key component for setting up CI/CD pipelines.
- B. Incorrect.
Incorrect: Amazon SageMaker Model Monitor is used for monitoring models in production, not for orchestrating or building CI/CD pipelines.
- C. Correct.
Correct: AWS CodeBuild is used for building and packaging code, which is an essential step in a CI/CD pipeline for ML workflows.
- D. Incorrect.
Incorrect: AWS Lambda is not typically used for deploying ML models; deployment is typically handled using Amazon SageMaker Endpoints or other services.
- E. Correct.
Correct: Amazon SageMaker Training Jobs can retrain the model, which is often a necessary step in CI/CD pipelines for machine learning.