Which AWS service enables modeling and prediction for device-generated data locally?

The correct answer is Amazon IoT Greengrass. This service is specifically designed to extend AWS functionality to edge devices that require local processing capabilities. With Greengrass, machine learning models can be deployed to devices, allowing them to make predictions locally without needing a constant cloud connection. This is particularly beneficial for processing device-generated data in real-time, especially in scenarios where low latency is crucial or where internet connectivity may be intermittent or unreliable.

Amazon IoT Core serves as a managed cloud service that allows devices to connect and interact with AWS services but does not provide the ability to execute machine learning models locally on those devices. Amazon Elastic Container Service (ECS) is used to run containerized applications in the cloud and is not focused on the local processing of data from edge devices. Amazon SageMaker is primarily a cloud-based machine learning platform used for building, training, and deploying models but does not natively offer capabilities for running predictions directly on edge devices. Thus, for modeling and prediction specifically designed to operate on device-generated data in a local environment, Amazon IoT Greengrass is the most suitable choice.