AWS re:Invent 2024 Recap: AI Part 2

Amazon SageMaker is a professional machine learning environment that supports training, building, and deploying AI models. Recently, through Hyperpod, it has further improved machine learning workflow efficiency by providing fully managed infrastructure with fault resilience and scaling.

Key updates are as follows:

• SageMaker Hyperpod flexible training plans: By entering compute requirements and training time windows, resources are automatically managed and clusters are created, lowering the difficulty of infrastructure management.
• SageMaker Hyperpod task governance: Coordinates resource allocation across departments and actively distributes idle resources, boosting resource utilization by up to 90%.
• Partner AI Apps: Embeds external AI development tools such as comet and deepchecks within SageMaker, allowing them to be used without security or operational burden.

Amazon Bedrock connects APIs of trained models, enabling even non-experts to easily develop AI applications. This provides an environment for quickly integrating generative AI features into services without a complex development process.