AWS Certified AI Practitioner Practice Exam

What is the best solution for a company to deploy an ML model for image classification without managing infrastructure?

1. A. Use Amazon SageMaker Serverless Inference to deploy the model

2. B. Use Amazon CloudFront to deploy the model

3. C. Use Amazon API Gateway to host the model and serve predictions

4. D. Use AWS Batch to host the model and serve predictions

The correct answer is: A. Use Amazon SageMaker Serverless Inference to deploy the model

The most suitable solution for deploying a machine learning model for image classification without the need to manage infrastructure is through Amazon SageMaker Serverless Inference. This option allows organizations to focus on their model and predictions rather than the underlying infrastructure. Amazon SageMaker is a fully managed service that facilitates the building, training, and deployment of machine learning models. The Serverless Inference feature specifically enables users to deploy their models in a serverless architecture, which automatically handles the scaling and infrastructure requirements. This means that you don't need to provision or manage any servers; the service can accommodate varying request loads seamlessly. It is designed to be cost-effective as well, as you pay only for the inference requests made when the model is used. Other options presented, such as using CloudFront or API Gateway, while useful for certain tasks, do not provide the necessary machine learning-specific features and capabilities for model deployment. CloudFront is primarily a content delivery network (CDN) and is not tailored for machine learning model inference. API Gateway can expose APIs but requires additional steps to integrate with ML models and manage the underlying infrastructure unless paired with a compute service, thus not being inherently serverless in the same way. AWS Batch is designed for high-throughput batch processing and is better