MLA-C01 Question 262
Select 2You are working as an ML Engineer at a healthcare company and are tasked with deploying a machine learning model to predict patient readmission probabilities. The model needs to serve predictions in two different ways: (1) in real-time for hospital staff accessing patient dashboards and (2) in batch for overnight processing of all patients in the system. Which of the following AWS services or solutions would you use to meet both requirements?
- A
Amazon SageMaker real-time endpoint for real-time predictions and AWS Glue for batch processing
- B
Amazon SageMaker real-time endpoint for real-time predictions and Amazon SageMaker Batch Transform for batch processing
- C
Amazon API Gateway with an Amazon SageMaker real-time endpoint for real-time predictions and AWS Batch for batch processing
- D
Amazon SageMaker multi-model endpoint for both real-time and batch processing
- E
Amazon Lambda for real-time predictions and AWS Step Functions for orchestrating batch jobs
Show answer and explanation
Correct answers: B, C
Explanation
To meet the requirements of serving both real-time and batch predictions, you can use Amazon SageMaker real-time endpoints for the real-time prediction needs and either SageMaker Batch Transform or AWS Batch for batch processing. Both solutions (options 2 and 3) are specifically designed to handle these workloads efficiently and are well-integrated with other AWS services. The other options either misuse services or are not optimized for the described use case.
- A. Incorrect.
Amazon SageMaker real-time endpoints are ideal for hosting models to serve real-time predictions. However, AWS Glue is primarily used for data integration tasks and ETL workflows, not for batch inference.
- B. Correct.
Amazon SageMaker real-time endpoints can serve real-time predictions while SageMaker Batch Transform is specifically designed for batch inference, making this a valid solution for the given requirements.
- C. Correct.
Amazon API Gateway can integrate with a SageMaker real-time endpoint for real-time predictions, and AWS Batch is a valid solution for running batch jobs, including batch inference.
- D. Incorrect.
Amazon SageMaker multi-model endpoints are optimized for hosting multiple models on a single endpoint and serving real-time requests, but they are not typically used for batch inference.
- E. Incorrect.
Amazon Lambda is capable of serving real-time predictions for lightweight models but is not the most efficient option for large-scale real-time inferencing. AWS Step Functions are used for orchestrating workflows but do not handle inference directly.