All news with #amazon sagemaker ai tag

🚀 Amazon SageMaker Feature Store now supports high-throughput ingestion and record discovery with new APIs and offline-store naming options. Data scientists can use BatchWriteRecord to write multiple records across feature groups in one request, and ListRecords to page through record identifiers without prior knowledge. Offline-store configuration now allows creating Glue and Iceberg tables with custom names, simplifying cataloging and management.

🛡️ AWS Config has expanded its set of managed rules with 191 additional checks covering services such as Amazon Bedrock, SageMaker, ECS, EKS, RDS, Redshift, S3, and CloudTrail. The new rules evaluate encryption, logging, public access, network security, data protection, and operational best practices. You can deploy rules individually or as part of a conformance pack in supported AWS Regions.

📊 Amazon SageMaker Unified Studio now supports OpenLineage-compatible data lineage in IAM-based domains, capturing events from Apache Spark on Amazon EMR, AWS Glue, SageMaker Visual ETL, and notebooks. The interactive lineage graph shows data flow with configurable depth, timestamp modes for column-level detail, and a dataset-only view. You can programmatically publish, query, manage, and delete lineage events via OpenLineage APIs and the DeleteLineageEvent API.

🚀 Amazon SageMaker Unified Studio can now connect to existing Amazon Managed Workflows for Apache Airflow (MWAA) environments, enabling teams to manage Airflow workflows from within their Studio projects. To add a connection, open the Workflows tool, choose "Add connection," and supply the Airflow configuration referencing your domain and project. Once connected, project members can sync, trigger, and monitor workflows; environments running Apache Airflow 3+ also gain the visual drag-and-drop authoring experience.

🚀 Amazon SageMaker HyperPod now supports Disaggregated Prefill and Decode (DPD), which splits LLM inference into separate prefill and decode GPU pools and transfers KV cache over EFA using GPU-Direct RDMA. This reduces contention between long-context prefill and per-token decode, enabling more consistent per-token latency, higher goodput under strict latency SLOs, and independent scaling of prefill and decode. DPD is enabled via a pdSpec in the existing InferenceEndpointConfig and works with HyperPod's KV cache offloading and intelligent routing on EFA-capable instances.

🔧 Amazon SageMaker Studio now offers direct, one‑click integration with Hugging Face so users can open a fully configured Studio environment with a selected model preloaded. Previously, users had to navigate console menus, configure IAM and serverless settings, and request GPU quotas; the new flow automates environment creation and permission configuration. New customers receive a Studio environment on sign‑up; verified customers gain default GPU access and visible quota information within Studio.

🔧 Amazon SageMaker Unified Studio now supports Terraform provisioning via the open-source terraform-aws-sagemaker-unified-studio module. Platform teams can deploy domains through version-controlled templates and integrate SageMaker Unified Studio into existing infrastructure-as-code pipelines. The module manages domain infrastructure and IAM roles, includes sub-modules for blueprints and projects, and uses the Terraform AWS Cloud Control Provider.

🛠️ Amazon SageMaker HyperPod now reports AMI versions across clusters and can automatically apply backward-compatible security patches without disrupting workloads. Administrators can view AMI semantic versions (major.minor.patch), detect drift, and roll back to prior versions — preserving NVIDIA drivers, CUDA, and other bundled software — via the UpdateClusterSoftware API. Auto-patching is opt-in per instance group, applies only when nodes are idle, and avoids major/minor upgrades; it can be enabled through CreateCluster or UpdateCluster APIs. A new AMI support policy defines patch support timelines; both features are available for EKS-orchestrated HyperPod clusters in supported Regions.

🛡️ Amazon Managed Service for Prometheus is now authorized for FedRAMP High and DoD CC SRG IL-4 and IL-5 in the AWS GovCloud (US) Regions. The fully managed, Prometheus-compatible monitoring service supports high-cardinality workloads with automatic scaling of ingestion and storage. It integrates with AWS security services to provide secure access to telemetry and alerting for regulated federal and public sector environments. Customers with FedRAMP High or DoD IL-4/5 requirements can adopt the service with confidence.

🚀 Amazon SageMaker Inference now supports container image caching to reduce scale-out latency for generative AI models. The service pre-caches the specified container image so new instances can begin serving without waiting to pull large images from Amazon ECR. This feature works with accelerator instance types, single-model endpoints, and inference component-based endpoints and requires no customer changes.

🛡️ AWS Security Hub CSPM introduces the AI Security Best Practices standard, offering 31 automated controls to detect misaligned AI resources. The standard evaluates Amazon Bedrock, Bedrock AgentCore, and Amazon SageMaker workloads against recommended configurations without manual rule creation. It covers domains like network isolation, encryption, VPC placement, KMS usage, private registries, and authorization, producing findings to help teams remediate issues. Available in all Regions where Security Hub CSPM operates, including GovCloud (US) and China.

🧰 Amazon SageMaker AI now supports serverless model customization for Gemma 4 E4B and 31B models using supervised fine-tuning (SFT), direct preference optimization (DPO), and reinforcement fine-tuning (RFT). You can adapt these Google DeepMind-built Gemma models to specific domains and workflows, and SageMaker AI handles infrastructure provisioning and training orchestration so teams pay only for what they use. The launch expands serverless customization to include models from Nova, Nemotron 3, Qwen, Llama, gpt-oss, and DeepSeek families, and is available in US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and EU (Ireland).

🔔 AWS is announcing availability changes affecting several services and features, with some moving to Maintenance and others entering Sunset or already at End of Support. Services moving to maintenance will not be accessible to new customers after specified dates, but existing customers can continue use. Several SageMaker features and select media and communication components are impacted. AWS provides migration guides and support resources to help affected customers.

🔍 Amazon Web Services announced the availability of all-MiniLM-L12-v2 in SageMaker JumpStart, expanding model options for customers. The Sentence Transformers model encodes sentences and short paragraphs into 384-dimensional dense vectors, enabling semantic search, clustering, and similarity tasks. Its compact architecture offers fast inference and strong embedding quality, suitable for production-scale text representation workloads.

🧭 AWS now offers Ministral-3-14B-Instruct-2512 in Amazon SageMaker JumpStart, adding a compact multimodal foundation model optimized for edge deployment. The 14B-parameter model supports image analysis, agentic workflows with native function calling and JSON output, and multilingual understanding across dozens of languages. Customers can deploy the model from SageMaker Studio or via the SageMaker Python SDK with a few clicks to build AI assistants, agentic systems, and vision-enabled applications on AWS.

🔍 Amazon SageMaker AI now provides built-in observability for production generative AI inference, delivering real-time visibility into token performance, GPU health, inference component placement, and autoscaling behavior. The new SageMaker AI Insights dashboard in Amazon CloudWatch surfaces metrics like Time to First Token, inter-token latency, queue depth, and tokens per second alongside infrastructure health, with OpenTelemetry native metrics published automatically. Customers using Grafana can connect via a regional PromQL endpoint and import a pre-configured dashboard template for unified monitoring and faster diagnosis.

🧭 AWS Glue Interactive Sessions now supports Apache Spark Connect, enabling development and execution of Spark applications from managed notebooks like Amazon SageMaker Unified Studio or IDEs such as Jupyter and VS Code while running on AWS Glue's serverless infrastructure. The thin client architecture decouples client dependencies from the server-side Spark runtime, enabling ad hoc exploration, iterative debugging, and incremental PySpark development. Observability includes real-time Spark UI monitoring, History Server tracking, and session management via the AWS Glue API, CLI, or SDK. This capability is available across multiple AWS regions.

🆕 Amazon S3 now supports annotations, enabling attachment of custom JSON, XML, or YAML metadata to objects at up to 1GB per object to provide business context for AI agents and analytics tools. Annotations persist with objects through replication and copy operations, can be modified or deleted at any time, and share the same durability and consistency as the object. You can surface annotations in S3 Metadata for query via Apache Iceberg tables, or search them with natural language using Amazon SageMaker Unified Studio and supporting tools.

🚀 Amazon SageMaker AI now supports serverless customization for Nvidia Nemotron 3 Nano via supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT). This open-weight 30B-parameter model can be deployed and adapted to specific domains and workflows directly within SageMaker. Serverless customization handles infrastructure and training orchestration, enabling teams to focus on data and evaluation while paying only for usage. The feature is available in US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and Europe (Ireland), and can be launched from SageMaker Studio or via the SageMaker Python SDK.

🚀 Amazon SageMaker Unified Studio Notebooks now support Amazon EMR Serverless with Apache Spark Connect, enabling PySpark and Spark SQL execution on EMR Serverless Spark Applications directly from notebook cells. Users can choose their Spark runtime from the Notebook side panel, with the selection applying to both Python and SQL cells. The feature integrates with SageMaker Data Agent to generate code and execution plans from natural language prompts and offers pre-initialized capacity, unified Spark UI monitoring, and VPC connectivity for isolated workloads. It is available in all regions where SageMaker Unified Studio is offered and supports both Unified Studio notebooks and JupyterLab IDEs.