MLS-C01 Question 293
Select 2You are tasked with deploying a machine learning model to production on AWS. The model requires low latency predictions and will be handling real-time inference requests. Following AWS best practices, which of the following actions should you take when deploying the model?
- A
Deploy the model using Amazon SageMaker Hosting Services with Auto Scaling enabled.
- B
Store the model artifacts in Amazon S3 and load them directly into the application at runtime.
- C
Enable Amazon CloudWatch metrics and logs to monitor the model's performance in production.
- D
Deploy the model on an Amazon EC2 instance and manually configure the environment for scaling.
- E
Use Amazon SageMaker Edge Manager to deploy the model for inference on IoT devices.
Show answer and explanation
Correct answers: A, C
Explanation
Following AWS best practices for deploying machine learning models, using Amazon SageMaker Hosting Services with Auto Scaling ensures scalable and low-latency real-time inference. Additionally, enabling Amazon CloudWatch metrics and logs provides critical monitoring capabilities, ensuring the model performs reliably in production. These practices minimize operational overhead and adhere to AWS's recommended deployment strategies.
- A. Correct.
Correct: Deploying the model using Amazon SageMaker Hosting Services with Auto Scaling ensures low-latency predictions and scalable infrastructure to handle real-time inference requests, aligning with AWS best practices.
- B. Incorrect.
Incorrect: Loading model artifacts from Amazon S3 into the application at runtime is inefficient for real-time inference and does not follow best practices for low-latency predictions.
- C. Correct.
Correct: Enabling Amazon CloudWatch metrics and logs allows for monitoring and troubleshooting the model's performance, which is a best practice for production deployments.
- D. Incorrect.
Incorrect: Deploying the model on an Amazon EC2 instance with manual scaling requires significant operational overhead and does not align with AWS best practices for scalability and efficiency.
- E. Incorrect.
Incorrect: Amazon SageMaker Edge Manager is designed for deploying models to edge or IoT devices, not for real-time inference in a production environment.