MLA-C01 Question 236
Select 3You are deploying a machine learning model using Amazon SageMaker and want to automate the process of retraining and redeploying the model whenever new data is available in an S3 bucket. Additionally, you want to ensure the workflow is orchestrated to include pre-processing, training, and deployment steps. Which combination of solutions will best meet your requirements?
- A
Use Amazon SageMaker Pipelines to orchestrate the ML workflow, including pre-processing, training, and deployment steps.
- B
Set up an AWS Lambda function to trigger the workflow whenever new data is uploaded to the S3 bucket.
- C
Use Amazon SageMaker Model Monitor to automatically retrain the model when new data becomes available.
- D
Integrate Amazon EventBridge with SageMaker Pipelines to trigger the workflow based on S3 events.
- E
Manually retrain the model using SageMaker Notebook Instances whenever new data is available.
Show answer and explanation
Correct answers: A, B, D
Explanation
The best solution involves using Amazon SageMaker Pipelines to orchestrate the ML workflow, AWS Lambda to trigger the workflow when new data is uploaded to the S3 bucket, and Amazon EventBridge to enable event-driven automation. These services together create a fully automated pipeline for retraining and redeploying the model. Amazon SageMaker Model Monitor and manual retraining do not fulfill the requirements for automation and orchestration in this scenario.
- A. Correct.
Correct: Amazon SageMaker Pipelines is designed to orchestrate the entire ML workflow, including pre-processing, training, and deployment steps. This is a key component for automating the ML lifecycle.
- B. Correct.
Correct: AWS Lambda can be used to trigger workflows based on events, such as new data being uploaded to an S3 bucket. This ensures automation of the retraining process.
- C. Incorrect.
Incorrect: Amazon SageMaker Model Monitor is used for monitoring deployed models for data drift, bias, or other issues. It does not handle retraining models automatically.
- D. Correct.
Correct: Amazon EventBridge can be configured to detect events (e.g., S3 object uploads) and trigger actions, such as starting a SageMaker Pipeline. This integrates well with S3 event-driven workflows.
- E. Incorrect.
Incorrect: Manually retraining the model is not an automated solution and does not align with the requirement to automate the retraining and redeployment process.