- SageMaker Data Wrangler is a new feature that promises to make data preparation easier. Data Wrangler is a point-and click interface that allows developers to quickly select the data from their AWS environment that they want to use in their machine learning model. They can then import it and prepare it with 300 pre-included transforms.
- SageMaker Clarify: Alerts developers about statistical biases in machine learning models.
- SageMaker Pipelines is a CI/CD service that is specifically designed for machine learning workflows.
- SageMaker EdgeManager: Provides machine learning model management and a speed boost for edge devices. This includes robots, smart cameras, smartphones, and computers.
- SageMaker Feature store: A repository to store, update and manage machine learning features. This repository is located directly in SageMaker so developers don’t have to use their Amazon S3 storage to manage their machine learning models.
- SageMaker JumpStart – A collection of pre-trained machine-learning solutions that developers can deploy right away or modify as needed.
- SageMaker Distributed Learning: This new capability promises to drastically reduce the time required to train large machine-learning models by automatically splitting them across multiple GPU instances.
- SageMaker Model Monitor improvements: This product alerts developers to any changes or drifts in their data over time. Model Monitor now allows you to check for model quality, bias, and feature importance.
- SageMaker Debugger improvements: Debugger can be used to monitor model training metrics in real-time so developers can fix issues sooner than later. The ability to monitor “system resources” such as CPU, GPU and network I/O, is a new feature of debugger.
All of the new features mentioned above are now generally available in select AWS regions.