Skip to content Skip to footer

Amazon SageMaker now collaborates with Amazon DataZone to enhance the management of machine learning processes.

Amazon has announced an integration between Amazon SageMaker, a fully managed machine learning (ML) service, and Amazon DataZone, a data management service. This integration is planned to facilitate infrastructure setup with security controls, collaboration on ML projects, and management of access to data and ML assets.

When solving business issues using ML, one creates ML models using training data, integrates them into business applications, and makes educated decisions based on the models. Effective ML governance while deploying such models helps establish trust in ML-powered applications, minimizes risks, and fosters responsible AI use.

However, applying governance across an ML lifecycle is challenging and time-consuming, as it usually requires custom workflows and the integration of multiple tools. This newly built integration between SageMaker and Amazon DataZone aims to simplify ML governance, collaboration on business initiatives, and control over data and ML assets in just a few clicks.

The integration provides capabilities in business project management, infrastructure management, and asset governance. Users can create, edit, and view projects; create multiple project environments and deploy infrastructure resources with embedded security controls; and manage access, discover, and publish data/ML assets to the enterprise business catalog.

Additionally, administrators and data stewards can manage and regulate access to data assets with Amazon DataZone. The service empowers various types of users such as engineers, data scientists, product managers, analysts, and business users to access data across the organization and derive insights.

This integration also introduces additional asset types such as SageMaker Feature Groups and Model Package Groups. Users can also manage and govern access to these subscribed assets and secured their data.

This integration should streamline ML governance, facilitate collaboration on shared business objectives, and boost control over data and ML assets, which will prove beneficial to data scientists and practitioners. Users can also quickly set up infrastructure resources with embedded security controls that cater to enterprise needs.

About the authors: Siamak Nariman is a Senior Product Manager at AWS, Kareem Syed-Mohammed is a Product Manager at AWS, Dr. Sokratis Kartakis is a Principal Machine Learning and Operations Specialist Solutions Architect at AWS, and Ram Vittal is a Principal ML Solutions Architect at AWS.

Leave a comment

0.0/5