Amazon

Innovating with machine learning on AWS

On AWS, you can access performant infrastructure, deployment resources, data governance solutions, and AWS Deep Learning Containers (DLCs) with optimized open source frameworks, so that you can focus on your machine learning tasks.

Build and Scale AI/ML on AWS

AWS offers a comprehensive suite of AI/ML tools and services supporting every stage of the machine learning lifecycle.Amazon SageMaker AI provides fully managed end-to-end model development and deployment, withSageMaker HyperPod enabling distributed training at scale featuring checkpointless recovery and elastic scaling.Amazon Bedrock enables building generative AI and agentic applications with leading foundation models including the Amazon Nova family (Nova 2 Sonic, Nova 2 Lite, Nova 2 Omni, Nova Act, and Nova Forge).Bedrock AgentCore rovides policy controls, quality evaluations, and enhanced memory for deploying trusted AI agents, while the open source while the open source Strands Agents SDK simplifies agent orchestration. Train your own models or fine-tune open source models cost-effectively using the new Trainium3 UltraServers powered by AWS's first 3nm AI chip. AWS custom AI accelerators also include Inferentia for low-latency inference and Graviton5 for best-in-class price performance. Whether you're building large language models, generative AI, AI agents, or computer vision applications—scaling on AWS is easy.

Learn more about these services and others:

• AWS GPUs for Machine Learning
• AWS Deep Learning AMIs
• Artificial Intelligence on AWS

Building AI Agents with Strands Agents SDK

AWS offers Strands Agents, an open source SDK that simplifies the development and deployment of AI agents. Supporting models from Amazon Bedrock, Meta's Llama, and other providers — Strands Agents enables developers to build production ready multi-agent systems with minimal code.

• Introduction to Strands Agents Blog
• Strands Agents on GitHub

AWS & Hugging Face Collaboration

AWS and Hugging Face are working together to simplify and accelerate the adoption of advanced machine learning models. This collaboration offers streamlined training using Hugging Face Deep Learning Containers with SageMaker AI distributed training libraries, simplifying workflows with the SageMaker AI Python SDK for efficient model training. Deployment is made effortless through the Hugging Face Inference toolkit and DLCs, allowing users to deploy trained models on the Hugging Face Hub. Amazon SageMaker AI facilitates the creation of scalable endpoints with built-in monitoring and enterprise-level security. This joint effort empowers teams to move quickly from experimentation to production, leveraging cutting-edge models and scalable infrastructure to drive innovation in machine learning projects.

• Learn about Hugging Face on AWS
• Reference documentation for using Hugging Face with Amazon SageMaker AI
• AWS Whitepaper- Scale production AI with Hugging Face Optimum Neuron and AWS Trainium and Inferentia
• Hugging Face Hub on AWS Marketplace

Connect, Learn, and Grow with AWS

Stay connected with the latest AWS AI/ML and open source developments:

Let's innovate together! 🎉🚀

Building intelligent physical AI from edge to cloud

Checkpointless training on Amazon SageMaker

Strands agent SOPS natural language workflows

Easy BYOC on SageMaker