Complete ML platform with Amazon SageMaker for building, training, and deploying machine learning models at scale
In this post, we demonstrate how Notebooks in Amazon SageMaker Unified Studio help you get to insights faster by simplifying infrastructure configuration. You'll see how to analyze housing price data, create scalable data tables, run distributed profiling, and train machine learning (ML) models within a single notebook environment.
In this post, you will learn how speculative decoding works and why it helps reduce cost per generated token on AWS Trainium2.
We're excited to announce the launch of Amazon SageMaker JumpStart optimized deployments. SageMaker JumpStart improved deployments address the need for rich and straightforward deployment customization on SageMaker JumpStart by offering pre-defined deployment configurations, designed for specific use cases. Customers maintain the same level of visibility into the details of their proposed deployments, but now deployments are optimized for their specific use case and performance constraint.
In this blog post, we use Athena and Amazon SageMaker Unified Studio to explore Parquet Column Indexes and demonstrate how they can improve Iceberg query performance. We explain what Parquet Column Indexes are, demonstrate their performance benefits, and show you how to use them in your applications.
Amazon SageMaker HyperPod task governance now supports gang scheduling, which ensures all pods required for a distributed training job are ready before training begins. Administrators can configure gang scheduling to prevent wasted compute from partial job runs and avoid deadlocks from jobs waiting for resources. Data scientists running distributed AI/ML training jobs on Amazon SageMaker HyperPod clusters using the EKS orchestrator require multiple pods to work together across nodes with pod-to-pod communication. When some pods start but others do not, jobs can hold onto resources without making progress, block other workloads, and increase costs. Gang scheduling resolves this by monitoring all pods in a workload and pulling the workload back if not all pods are ready within a set time. Pulled-back workloads are automatically requeued to prevent stalling. Administrators can adjust settings on the HyperPod Console, such as how long to wait for pods to be ready, how to handle node failures, whether to admit workloads one at a time to avoid deadlocks on busy clusters, and how retries are scheduled. This capability is currently available for Amazon SageMaker HyperPod clusters using the EKS orchestrator across the following AWS Regions: US East (N. Virginia), US East (Ohio), US West (N. California), US West (Oregon), Asia Pacific (Mumbai), Asia Pacific (Singapore), Asia Pacific (Sydney), and Asia Pacific (Tokyo), Asia Pacific (Jakarta), Europe (Frankfurt), Europe (Ireland), Europe (London), Europe (Stockholm), Europe (Spain), and South America (São Paulo). To learn more, visit SageMaker HyperPod webpage, and HyperPod task governance documentation.
Amazon SageMaker Unified Studio now supports Serverless Workflows in Identity Center domains. With this launch, customers using Identity Center domains can orchestrate data processing tasks with Apache Airflow (powered by Managed Workflows for Apache Airflow) without provisioning or managing Airflow infrastructure. Serverless Workflows were previously available only in IAM-based domains. Serverless Workflows automatically provision compute resources when a workflow runs and release them when it completes, so you only pay for actual workflow run time. Each workflow runs with its own execution role and isolated worker, providing workflow-level security and preventing cross-workflow interference. With Serverless Workflows, Identity Center domain customers also get access to the Visual Workflow experience with support for around 200 operators, including built-in integration with AWS services such as Amazon S3, Amazon Redshift, Amazon EMR, AWS Glue, and Amazon SageMaker AI. Serverless Workflows in Identity Center domains are available in all AWS Regions where SageMaker Unified Studio is supported. To learn more, visit the Serverless Workflows documentation.
In this post, we walk through the new installation experience, demonstrate three deployment methods (console, CLI, and Terraform), and show how features like multi-instance-type deployment and native node affinity give you fine-grained control over inference scheduling
Amazon SageMaker Unified Studio notebooks now support import/export capabilities, enabling migration from JupyterLab and other notebook platforms. This release also introduces developer acceleration features including cell reordering, keyboard shortcuts, cell renaming, and multi-line SQL support, designed to enhance productivity for data engineers and data scientists professionals working with notebook-based workflows. The new import/export functionality supports .ipynb, .json, and .py formats while preserving cell types, metadata, and outputs, making platform migration straightforward. You can export notebooks in four formats including Jupyter notebook with requirements (.zip), standard .ipynb, Python scripts (.py), and SageMaker Unified Studio native format (.json). Developer acceleration features enable you to reorder cells without copy-paste duplication, assign custom names to cells for improved navigation in large notebooks, use familiar keyboard shortcuts for faster development, and execute multiple SQL statements in a single cell with results displayed in separate tabs for easy comparison and analysis. This feature is available in all AWS Regions where Amazon SageMaker Unified Studio is available. To learn more, visit the Amazon SageMaker Unified Studio marketing page and user guide.
Today, AWS announces the launch of Partner Revenue Measurement integration with AWS Marketplace Metering for Amazon Machine Image (AMI) and Machine Learning (ML) products listed in AWS Marketplace. Partner Revenue Measurement allows Partners to better understand their AWS revenue impact and product consumption patterns. The AWS Marketplace Metering capability automatically measures AWS service consumption when customers purchase and use AMI and ML products via AWS Marketplace. Partners can now gain visibility into how their solutions impact Amazon Elastic Compute Cloud (Amazon EC2) and Amazon SageMaker AI service consumption across partner-managed and customer-managed accounts. This method complements Partner Revenue Measurement’s Resource Tagging and User Agent string capabilities by capturing attribution without requiring additional Partner implementation. Partner Revenue Measurement is generally available in all commercial regions. To learn more about AWS Marketplace Metering, review the AWS Marketplace metering guide. To learn more about Partner Revenue Measurement capabilities, review the onboarding guide.
Amazon CloudWatch introduces Container Insights with OpenTelemetry metrics for Amazon EKS, available in public preview. Building on the existing Container Insights experience, this capability provides deeper visibility into EKS clusters by collecting more metrics from widely adopted open source and AWS collectors and sending them to CloudWatch using the OpenTelemetry Protocol (OTLP). Each metric is automatically enriched with up to 150 descriptive labels, including Kubernetes metadata and customer-defined labels such as team, application, or business unit. Curated dashboards in the Container Insights console present cluster, node, and pod health with the ability to aggregate and filter metrics by instance type, availability zone, node group, or any custom label. For deeper analysis, customers can write queries using the Prometheus Query Language (PromQL) in CloudWatch Query Studio. The CloudWatch Observability EKS add-on provides one-click installation through the Amazon EKS console, or can be deployed through CloudFormation, CDK, or Terraform. The add-on automatically detects accelerated compute hardware including NVIDIA GPUs, Elastic Fabric Adapters, and AWS Trainium and Inferentia accelerators. For existing customers of the add-on, CloudWatch supports publishing both OpenTelemetry and existing Container Insights metrics at the same time. Container Insights with OpenTelemetry metrics is available in public preview in US East (N. Virginia), US West (Oregon), Asia Pacific (Sydney), Asia Pacific (Singapore), and Europe (Ireland). There is no charge for OpenTelemetry metrics from Container Insights during preview. To get started, see the Container Insights with OpenTelemetry metrics for Amazon EKS.
In this post, we explore scenarios where customers need more control over their network infrastructure when building their unified data and analytics strategic layer. We’ll show how you can bring your own Amazon Virtual Private Cloud (Amazon VPC) and set up Amazon SageMaker Unified Studio for strict network control.
In this post, we explore SageMaker Unified Studio multi-account deployments in depth: what they entail, why they matter, and how to implement them effectively. We examine architecture patterns, evaluate trade-offs across security boundaries, operational overhead, and team autonomy. We also provide practical guidance to help you design a deployment that balances centralized control with distributed ownership across your organization.
In this post, you will learn how Aigen modernized its machine learning (ML) pipeline with Amazon SageMaker AI to overcome industry-wide agricultural robotics challenges and scale sustainable farming. This post focuses on the strategies and architecture patterns that enabled Aigen to modernize its pipeline across hundreds of distributed edge solar robots and showcase the significant business outcomes unlocked through this transformation. By adopting automated data labeling and human-in-the-loop validation, Aigen increased image labeling throughput by 20x while reducing image labeling costs by 22.5x.
This post is co-written with Neel Patel, Abdullahi Olaoye, Kristopher Kersten, Aniket Deshpande from NVIDIA. Today, we’re excited to announce that the NVIDIA Evo-2 NVIDIA NIM microservice are now listed in Amazon SageMaker JumpStart. You can use this launch to deploy accelerated and specialized NIM microservices to build, experiment, and responsibly scale your drug discovery […]