MLA-C01 Question 261
Select 2You are a Machine Learning Engineer tasked with deploying a trained model for a retail application. The model needs to provide real-time predictions for customer queries via an API and also periodically process large datasets for generating daily sales forecasts. Which combination of AWS services would best meet these requirements?
- A
Amazon SageMaker Hosting Endpoints for real-time inference and AWS Batch for batch processing
- B
Amazon SageMaker Hosting Endpoints for real-time inference and Amazon SageMaker Processing Jobs for batch processing
- C
Amazon Elastic Inference for real-time inference and Amazon Athena for batch processing
- D
Amazon SageMaker Asynchronous Inference for real-time inference and AWS Glue for batch processing
- E
Amazon SageMaker Hosting Endpoints for real-time inference and AWS Glue for batch processing
Show answer and explanation
Correct answers: A, E
Explanation
Amazon SageMaker Hosting Endpoints are ideal for serving ML models in real-time via APIs, making them a good choice for customer query predictions. For batch processing, AWS Batch or AWS Glue can be used depending on the use case. AWS Batch is designed for batch computing workloads, while AWS Glue is tailored for ETL tasks and large-scale data processing. Both combinations (SageMaker Hosting Endpoints + AWS Batch and SageMaker Hosting Endpoints + AWS Glue) can effectively fulfill the requirements of the scenario.
- A. Correct.
Correct. Amazon SageMaker Hosting Endpoints are designed for real-time inference, while AWS Batch is optimized for processing large datasets in batch mode.
- B. Incorrect.
Incorrect. While Amazon SageMaker Hosting Endpoints can handle real-time inference, SageMaker Processing Jobs are better suited for data preprocessing or model evaluation tasks rather than large-scale batch inference.
- C. Incorrect.
Incorrect. Amazon Elastic Inference is a cost-effective way to add GPU acceleration to real-time inference but does not handle batch processing, and Amazon Athena is for querying data in S3 with SQL, not batch inference.
- D. Incorrect.
Incorrect. SageMaker Asynchronous Inference is designed for handling large payloads and delayed inference, not real-time predictions. AWS Glue is more suited for ETL tasks than batch inference.
- E. Correct.
Correct. Amazon SageMaker Hosting Endpoints are suitable for real-time inference, and AWS Glue can be used for batch processing workflows, such as generating daily sales forecasts by transforming and analyzing large datasets.