An AWS post demonstrates deploying Hugging Face models to SageMaker AI using six open-source agent skills from the Hugging Face Skills repo, working with Kiro and Claude Code. Unguided agents chose outdated TGI containers and produced failed or silently broken deployments; the skills select vLLM from the AWS DLC catalog and add autoscaling, alarms, and teardown.
AWS announced the aws-ai-ml skill, available through the Agent Toolkit for AWS, which gives coding agents like Kiro, Claude Code, and Codex SageMaker AI inference optimization expertise. Agents can benchmark live endpoints, recommend deployment configurations, compare benchmark runs, and generate executable SageMaker Python SDK v3 code. The skill works with any MCP-supporting agent and can also run in a pre-configured SageMaker Studio JupyterLab space.
AWS announced the aws-ai-ml skill, available through the Agent Toolkit for AWS, which gives coding agents like Kiro, Claude Code, and Codex SageMaker AI inference optimization expertise. Agents can benchmark live endpoints, recommend deployment configurations, compare benchmark runs, and generate executable SageMaker Python SDK v3 code. The skill works with any MCP-supporting agent and can also run in a pre-configured SageMaker Studio JupyterLab space.
An AWS engineering post walks through using SageMaker AI Inference Recommendations to run automated concurrency sweeps, benchmarking NVIDIA Nemotron-3 Nano 30B on ml.g7e.2xlarge instances to find the highest concurrency that stays within latency SLAs. The feature is built into SageMaker, so teams avoid building custom load-testing infrastructure, and a search recipe can auto-discover the saturation point cheaper than a linear sweep.
An AWS engineering post walks through using SageMaker AI Inference Recommendations to run automated concurrency sweeps, benchmarking NVIDIA Nemotron-3 Nano 30B on ml.g7e.2xlarge instances to find the highest concurrency that stays within latency SLAs. The feature is built into SageMaker, so teams avoid building custom load-testing infrastructure, and a search recipe can auto-discover the saturation point cheaper than a linear sweep.
A workflow for deploying pretrained models from Amazon SageMaker JumpStart and running lightweight, repeatable performance tests with the SageMaker Python SDK, then deleting the endpoint. It helps ML practitioners evaluate whether a model suits their application without building custom test infrastructure.
A workflow for deploying pretrained models from Amazon SageMaker JumpStart and running lightweight, repeatable performance tests with the SageMaker Python SDK, then deleting the endpoint. It helps ML practitioners evaluate whether a model suits their application without building custom test infrastructure.
Kodama is a collection of multi-agent configurations for Kiro, AWS's agentic IDE. It uses LLM agents that can be composed into teams to work on coding tasks together. It targets developers using Kiro who want pre-built agent setups instead of configuring agents from scratch.
Kodama is a collection of multi-agent configurations for Kiro, AWS's agentic IDE. It uses LLM agents that can be composed into teams to work on coding tasks together. It targets developers using Kiro who want pre-built agent setups instead of configuring agents from scratch.
An AWS tutorial showing how to deploy vLLM-Omni on SageMaker AI to run text-to-image generation followed by image-to-video animation. It uses two endpoints from one Deep Learning Container: a real-time endpoint for FLUX.2-klein-4B image generation and an asynchronous endpoint for Wan2.1-VACE-1.3B video generation. It solves the problem of deploying multimodal generative models on managed AWS infrastructure for ML engineers.
An AWS tutorial showing how to deploy vLLM-Omni on SageMaker AI to run text-to-image generation followed by image-to-video animation. It uses two endpoints from one Deep Learning Container: a real-time endpoint for FLUX.2-klein-4B image generation and an asynchronous endpoint for Wan2.1-VACE-1.3B video generation. It solves the problem of deploying multimodal generative models on managed AWS infrastructure for ML engineers.