40 articles tagged with “mlops”.
AWS presents a SageMaker SDK v3 reference architecture for fleet-wide edge computer vision, training and versioning per-camera models with Step Functions pacing retraining within account quotas.
AWS details a reusable CDK construct using CloudFormation custom resources to deregister and re-register SageMaker AI endpoint auto scaling, allowing endpoint and scaling updates in one deployment.
Amazon SageMaker AI now offers generally available training sessions for serverless model customization, letting users run custom post-training loops without managing GPUs.
AWS has introduced a new onboarding experience that lets new SageMaker AI customers reach a fully configured SageMaker Studio environment using an existing identity such as Amazon, Apple, GitHub, or Google.
A recap of Amazon SageMaker JumpStart's 2026 year-to-date progress, including over 100 new models from 30 providers and platform enhancements.
AWS outlines a reference architecture for sharing a single SageMaker HyperPod EKS cluster across teams, combining IAM Identity Center authentication, per-team isolation, task governance, and namespace-level chargeback.
SageMaker Studio now supports creating, starting, stopping, and opening SageMaker Spaces on HyperPod EKS clusters, letting users launch JupyterLab and Code Editor without command-line tools.
AWS presents an architecture combining Amazon EKS, EFA, and S3 to scale Mixture-of-Experts reinforcement learning, reporting a 40% increase in aggregate rollout throughput for RLHF and GRPO training.
AWS details how to use SkyRL on SageMaker HyperPod to post-train a Qwen3-VL-8B vision-language model with GRPO, covering setup, job execution, and adapter hosting.
BMW Group's CLEA FinOps platform gained automated daily cost anomaly detection across more than 14,000 cloud accounts using Prophet forecasting and a serverless AWS pipeline.
Posit's Positron IDE now runs on Amazon SageMaker AI, enabling data scientists to explore data, train models in R and Python, deploy endpoints, and report results in one governed environment.
NVIDIA frames AI security as an engineering discipline requiring defined requirements, enforceable controls, clear ownership and evidence, and calls for faster, broader defensive tooling.
AWS reviews 13 inference launches shipped year-to-date in 2026 for Amazon SageMaker AI, spanning managed endpoints and SageMaker HyperPod Inference.
AWS introduced SageMaker HyperPod Inference Gateway, a Kubernetes-native, GPU-aware routing add-on for Amazon EKS that routes inference requests using real-time GPU signals.
A technical guide to using NVIDIA's Resiliency Extension with PyTorch FSDP on Amazon EKS, covering async checkpointing and fast GPU fault recovery with H100 benchmarks.
Amazon SageMaker AI now lets users specify an ordered list of up to five instance types for training and processing jobs, automatically launching on the first available option.
SageMaker HyperPod now supports model caching for inference, pre-loading model weights and container images onto cluster nodes so pods read from local NVMe instead of downloading over the network.
This piece discusses the rebuilding of AUTOMATIC1111 using Gradio Workflow to enhance development processes.
Amazon SageMaker Feature Store introduced the UpdateRecord API, enabling updates to one or more feature values in a single call without reading or rewriting the full record.
AWS extends MLflow and SageMaker AI Model Registry sync to two cross-account governance patterns: a hub-and-spoke model centralized with AWS RAM and a hybrid model that isolates development accounts.
AWS introduced HyperPod InstantStart, an open source control plane combining Amazon EKS with SageMaker HyperPod to run guarded operations via a web interface and an AI agent.
Hugging Face has unveiled NeoMME, a new encoder designed for efficient multimodal and multilingual processing.
SageMaker HyperPod now offers managed Ray support on Amazon EKS, letting users create and monitor Ray clusters, connect notebooks, and run distributed training and accelerated inference.
An AWS blog post describes AI-powered approaches to metadata correction and harmonization, including human-in-the-loop validation and autonomous agent workflows, plus production governance considerations.
AWS walks through building a no-code fraud detection model by connecting SageMaker Canvas to Snowflake and training XGBoost without writing ML code.
The second post in an AWS multi-agent series examines patterns that let ML teams run many agentic AI systems across diverse frameworks, models, and providers without vendor lock-in.
Jumio built a centralized, real-time feature store on AWS delivering sub-100ms feature serving for fraud detection while saving roughly $120,000 annually.
A Hugging Face Blog post describes achieving 33 percentage points more utilization on the same cluster by changing the order rather than the hardware.
AWS details how Amazon Bedrock AgentCore Observability can track AI agents running outside AWS, routing traces, span metrics, and token usage to one dashboard.
This AWS guide explains how to set up CUR 2.0 with IAM principal data and use Amazon Athena and CUDOS dashboards to track and analyze Amazon Bedrock costs across an organization.
AWS describes how the SageMaker AI Spaces add-on for Amazon EKS runs managed JupyterLab and Code Editor environments on an existing cluster, with browser and VS Code access.
Nscale, a British AI neocloud, is acquiring Anyscale, a software startup that helps companies scale AI workloads across data centers and servers.
A Hugging Face Blog post addresses GPU management, framing idle GPUs as a costly problem akin to grounded aircraft.
AWS presents a solution for centralized monitoring of SageMaker Pipelines across AWS accounts and Regions using CloudWatch custom dashboards, backed by a CDK example on GitHub.
This article discusses the lessons learned from building the Shippy agent, focusing on design principles for effective AI agents.
The Hugging Face Blog discusses the intricacies of model routing, highlighting challenges that arise in AI systems and their impact on performance.
An AWS blog post explains four deployment patterns for serving Unsloth-quantized models on AWS infrastructure, covering EC2, SageMaker AI endpoints, EKS, and ECS.
The article delves into various profiling methods in PyTorch, emphasizing attention mechanisms in models to optimize performance.
This blog post addresses common mistakes in MCP tool design and offers practical strategies for effective context engineering.
AWS details five new inference capabilities for SageMaker HyperPod, including multi-tier data capture, direct Hugging Face Hub deployment, NVMe model loading, Route 53 DNS, and pod-level IAM.