Developers now have the option to use Amazon SageMaker's fully managed MLflow

Amazon recently announced the general availability of a fully managed MLflow capability on Amazon SageMaker. MLflow is a popular open-source MLOps platform enabling users to manage all aspects of the ML lifecycle. The managed MLflow in Sagemaker lets users easily and quickly set up and manage MLflow Tracking Servers to boost their productivity and streamline their workflows. Being a managed service, developers are spared from the chores related to setting up and managing MLflow and can begin to leverage secure and scalable MLflow environments quickly and efficiently. The managed MLflow in Sagemaker is based on three core components: an MLflow Tracking Server, a backend metadata store, and an artifact store.

The standalone Tracking Server serves the necessary REST API endpoints for run and experiment tracking, thus optimizing the process of monitoring ML experiments. The backend metadata store retains metadata related to experiments, runs, and artifacts to ensure comprehensive ML experiments management and tracking. Finally, the artifact store compiles the artifacts generated during ML experiments, including trained models, datasets, logs, and plots in a customer-managed Amazon Simple Storage Service (S3) bucket for secure and efficient storage.

In addition to saving developers' time, Amazon SageMaker with MLflow brings other benefits, including comprehensive experiment tracking, all the standalone MLflow capabilities, unified model governance, efficient server management, and more. Amazon SageMaker with MLflow is generally available in all AWS Regions with SageMaker Studio availability, except China and US GovCloud Regions.