Consider subscribing to our daily and weekly newsletters to stay updated with the latest in AI industry coverage and exclusive content.
While Amazon Bedrock has garnered a lot of attention in the AI world over the past year, Amazon SageMaker continues to be a crucial service on AWS. Launched in 2017, Amazon SageMaker manages the entire machine learning lifecycle—from building and training models to deploying and managing them at scale. It provides a managed environment and tools for customers to create, train, and deploy machine learning and deep learning models. Hundreds of thousands of customers use Amazon SageMaker for tasks such as training popular generative AI models and deploying machine learning workloads. Notably, it played a role in training Stability AI’s Stable Diffusion and enabled Luma’s Dream Machine text-to-video generator.
AWS is now enhancing these capabilities with the general availability of the managed MLflow on SageMaker service. MLflow is a popular open-source platform that handles the entire machine learning lifecycle, including experimentation, reproducibility, deployment, and monitoring. By making MLflow a managed service within SageMaker, AWS is providing users with more tools and options to build the next generation of AI models.
Ankur Mehrotra, the director and general manager of Amazon SageMaker at AWS, emphasized that customers are eager to quickly transition from experimentation to production to accelerate time to market. The new managed MLflow capability allows users to set up and launch MLflow within the SageMaker environment with just a few clicks.
MLflow is widely used by developers and organizations for MLOps. Mehrotra highlighted that the new managed MLflow on SageMaker offers enterprise users more choices without replacing existing features. By integrating MLflow as a fully managed service with SageMaker, AWS aims to provide a seamless experience that combines the strengths of both platforms.
Users can iterate over their models, log metrics in MLflow, and track and compare different versions effortlessly. They can also register models in a model registry and deploy them easily. The managed MLflow service is deeply integrated with SageMaker components and workflows, ensuring actions in MLflow automatically sync with services like the SageMaker Model Registry.
Several organizations, including web hosting provider GoDaddy and Toyota Connected, have already tested the managed service during its beta phase.
AWS also offers a complementary service, Amazon Bedrock, which focuses on building generative AI applications. Mehrotra explained that while SageMaker is for building, training, and deploying models, Bedrock is designed for creating generative AI-based applications. Many AWS customers use both SageMaker and Bedrock to develop their AI solutions, taking advantage of Bedrock’s serverless capabilities to deploy models created in SageMaker.
Looking ahead, the development of Amazon SageMaker will focus on improving scalability and optimizing costs. AWS aims to reduce the heavy lifting for customers, making it easier and faster for them to create and market new AI solutions. Expect to see more features from AWS that simplify the creation and deployment of AI solutions.
Stay informed with the latest news by subscribing to our daily newsletter.