All news with #amazon sagemaker ai tag

🔒 SageMaker Data Agent now supports cross-region inference profiles for Japan (JP-CRIS) and Australia (AU-CRIS) via Amazon Bedrock. Inference requests originating in Asia Pacific (Tokyo) and Asia Pacific (Sydney) are processed entirely within their respective geographies, helping customers meet data residency and sovereignty requirements. Data Agent continues to provide conversational data exploration, Python and SQL code generation, troubleshooting, and analytics inside SageMaker Unified Studio Notebooks and the Query Editor, with traffic routed exclusively over the AWS Global Network.

🔍 Amazon SageMaker Unified Studio now surfaces Amazon CloudWatch metrics for AWS Glue jobs directly alongside job logs in a single, unified interface. Data engineers can correlate DPU utilization, memory consumption, CPU load, and data movement size with log output to diagnose compute bottlenecks and memory pressure faster. The consolidated view reduces mean time to resolution for ETL pipelines and is available in all Regions where SageMaker Unified Studio is generally available. To view metrics, open a Glue job run and select the Metrics tab.

🔍Data Agent in the Amazon SageMaker Unified Studio Query Editor brings natural-language-to-SQL capabilities to your SQL analytics workflow. You can ask questions in plain language and have the agent generate context-aware SQL for Amazon Redshift and Amazon Athena, propose step-by-step plans, and use Fix with AI to diagnose and correct failed queries. It preserves query context across follow-ups and is available in IAM domains where SageMaker Unified Studio is supported.

🔗 AWS now enables remote connections from Kiro and Cursor IDEs to Amazon SageMaker Studio. Data scientists, ML engineers, and developers can use their local Kiro/Cursor setups — including spec-driven development, conversational coding, and automated feature generation — while running workloads on SageMaker's scalable cloud compute. Authentication is supported via the AWS Toolkit extension or SageMaker Studio's web UI, preserving Studio security boundaries and reducing context switching.

🔗 AWS announced remote connection from Cursor IDE to SageMaker Unified Studio via the AWS Toolkit extension. This integration lets data scientists, ML engineers, and developers use their local Cursor setup — including AI-powered code completion, natural language editing, and multi-file editing — while leveraging SageMaker's scalable compute and data. Authentication is handled securely through AWS IAM via the Toolkit, preserving custom Cursor settings and enterprise-grade security.

⚙️ AWS Batch now supports quota management with job preemption for SageMaker Training, enabling prioritized GPU allocation and automatic preemption of lower-priority workloads. You can create up to 20 quota shares per job queue and choose resource-sharing strategies, with both cross-share and in-share preemption modes to restore or reallocate borrowed capacity. Capacity utilization is visible at queue, quota share, and job levels, and you can update job priorities after submission and set preemption retry limits. The feature integrates with the SageMaker Python SDK via the aws_batch module and is available in all AWS Regions where AWS Batch is offered; AWS provides an example notebook and user-guide documentation for implementation guidance.

🚀 Amazon SageMaker AI now offers serverless model customization and reinforcement fine-tuning for 12 additional open‑weight models, enabling SFT, DPO, and advanced RFT techniques such as RLVR and RLAIF without infrastructure management. You can fine‑tune and evaluate these models on a pay‑per‑use basis across multiple regions. This simplifies alignment for complex, domain‑specific tasks and improves accuracy on verifiable tasks like code generation and structured extraction. No cluster setup, capacity planning, or distributed training expertise is required.

🚀 Amazon SageMaker HyperPod now supports continuous provisioning for clusters using the Slurm orchestrator, allowing training jobs to start immediately on available instances while remaining capacity is provisioned in the background. Priority-based provisioning brings up the Slurm controller first, then login and worker nodes in parallel, with asynchronous retries for failed launches. The feature reduces time-to-training, improves utilization, and removes the need for manual scaling interventions.

🔎 Amazon SageMaker Unified Studio now supports custom metadata search filters, enabling teams to narrow catalog results using organization-specific attributes like business region, data classification, or study name. Filters accept string fields with a contains operator and numeric fields (Integer, Long) with equals, greater than, and less than operators. Users can also filter by asset name, description, and date range, combine multiple filters, and retain selections across browser sessions; the feature is available in all AWS Regions where Unified Studio is supported.

🔍 Amazon SageMaker Unified Studio now provides an aggregated view of data lineage that consolidates all jobs contributing to a dataset. The aggregated view displays multi-level transformations and dependencies to help you identify upstream sources and downstream consumers across the full lineage graph. It is the default for IdC-based domains, with an option to revert to the previous event-timestamp-ordered view, and the new QueryGraph API exposes node graphs with metadata and augmented business context.

🔍 Amazon SageMaker Unified Studio now offers an aggregated lineage view that shows all jobs contributing to a dataset across multiple levels of the lineage graph. The aggregated view is the default for IdC-based domains, while the previous event-timestamp snapshot can be restored via a "display in event timestamp order" toggle. A new QueryGraph API returns lineage node graphs with metadata and augmented business context; the capability is available in all SageMaker Unified Studio regions, with documentation and API references provided.

🔁 Amazon SageMaker Training Plans can now be extended to cover AI training runs that take longer than originally scheduled. Extensions are available in 1-day increments up to 14 days, or 7-day increments up to 182 days, and can be purchased through the SageMaker console or via API. Once an extension is purchased the reserved GPU capacity (clusters up to 64 instances) continues to run without interruption, and SageMaker automatically provisions infrastructure so workloads keep running without reconfiguration. The feature helps teams maintain cost-efficient training schedules and reduce operational disruptions.

🔆 AWS has added light mode support to Amazon SageMaker Unified Studio for IAM-based domains, allowing users to choose between dark and light visual themes. The addition improves readability in bright environments and offers a familiar look for customers who prefer lighter interfaces. In Studio, select Profile > customize appearance to switch modes. The setting persists across browsers and devices and complements the existing dark mode to give users full control over their development environment's appearance.

🔍 Amazon SageMaker HyperPod now provides integrated observability for Restricted Instance Groups (RIG), giving teams training foundation models with Nova Forge a unified view of compute resources and training workloads. A pre-configured Amazon Managed Grafana dashboard, backed by Amazon Managed Service for Prometheus, aggregates metrics from four exporters to show GPU utilization, NVLink bandwidth, CPU pressure, FSx for Lustre usage, network fabric, Kubernetes state, and curated logs including epoch progress, step-level logs, pipeline errors, and Python tracebacks. Observability is automatically enabled for new RIG clusters and can be turned on for existing clusters via the HyperPod console; it is available in all Regions where SageMaker HyperPod RIG is supported.

🔁 Amazon SageMaker Unified Studio now synchronizes catalog metadata and context with Atlan, Collibra, and Alation, aligning projects, assets, descriptions, glossary terms, and hierarchies across platforms. Collibra supports bidirectional synchronization and can manage SageMaker Unified Studio data access requests, while Atlan and Alation ingest metadata from SageMaker with additional enhancements planned. The Collibra integration is provided as an open-source solution on GitHub, and setup is performed by establishing connections from each partner to SageMaker Unified Studio.

🚀 Amazon SageMaker Unified Studio now supports AWS Glue 5.1 for Visual ETL, notebook, and code-based data processing jobs. With Glue 5.1 you can run on Apache Spark 3.5.6 with Python 3.11 and Scala 2.12.18, and use updated open table formats including Apache Iceberg 1.10.0, Apache Hudi 1.0.2, and Delta Lake 3.3.2. Select Glue 5.1 from the job version dropdown to apply the runtime across Visual ETL, notebooks, and code jobs.

🔗 AWS now enables Kiro IDE to connect remotely to Amazon SageMaker Unified Studio, allowing data scientists, ML engineers, and developers to use their local Kiro setup — including spec-driven development, conversational coding, and automated feature generation — while running workloads on SageMaker’s scalable compute. The integration uses the AWS Toolkit extension for secure IAM-based authentication and preserves local specs, steering files, and hooks. This reduces context switching and keeps agentic development workflows within a single environment across AWS analytics and ML services. The capability is available in all Regions where SageMaker Unified Studio is offered.

🔧 Amazon SageMaker HyperPod now supports API-driven Slurm configuration, enabling you to define Slurm topology, instance group to partition mappings, and FSx filesystem mounts directly in the cluster CreateCluster and UpdateCluster APIs or via the AWS Console. The update lets you specify node roles such as Controller, Login, and Compute per instance group and mount FSx for Lustre or FSx for OpenZFS filesystems. A new SlurmConfigStrategy (Managed, Overwrite, Merge) detects partition-node drift and controls whether updates are paused, overwritten, or merged to preserve manual customizations.

🔧 Amazon SageMaker HyperPod now lets operators manage individual cluster nodes directly from the AWS Console. The console enables SSM session launches, copyable pre-populated SSM CLI commands, and direct node actions such as reboot, delete, and replace, with support for batch operations across multiple nodes. Available in all Regions where HyperPod is supported, these controls reduce context switching and speed manual recovery for time-sensitive AI training and inference workloads.

🔈 Cartesia's Sonic 3 model is now available in Amazon SageMaker JumpStart, giving AWS customers a turnkey option for advanced streaming text-to-speech. Sonic 3 is a state space model (SSM) offering high naturalness, accurate transcript following, sub-100ms latency, and fine-grained control over volume, speed, and emotion. It supports 42 languages, natural laughter, and voices optimized for agents and expressive characters. Deployments can be launched from SageMaker Studio or via the SageMaker Python SDK.