From Hugging Face to Amazon SageMaker Studio in one click
23:15 · July 7, 2026 · Hugging Face Blog

Summary
Hugging Face and Amazon Web Services have introduced a one-click integration that lets users move directly from a model page on the Hugging Face Hub to Amazon SageMaker Studio for customization or deployment. The feature addresses the previous requirement to open the AWS console, create a SageMaker domain, set up IAM roles, and separately request GPU quota before any work could begin.
The integration adds three concrete capabilities. Model pages on Hugging Face now display buttons labeled “Customize on SageMaker AI” and “Deploy on SageMaker AI” for supported models; selecting either option signs the user into the console and lands them on the corresponding Studio page with the model already selected. New Studio environments receive the managed policy AmazonSageMakerModelCustomizationCoreAccess, which grants permissions for serverless jobs that use supervised fine-tuning, direct preference optimization, reinforcement learning with verifiable rewards, and reinforcement learning from AI feedback, along with deployment to SageMaker or Bedrock endpoints. The Studio interface also displays current quota availability for GPU instance types such as G5 and G6 directly in the instance-selection list, eliminating a separate visit to the Service Quotas console.
These changes preserve model context across the hand-off and remove manual configuration steps for both new and existing Studio environments. The result is a shorter path from model discovery to fine-tuning or inference within an enterprise AWS account.
Why it matters
This article is highly relevant for ML Engineers as it introduces a streamlined MLOps workflow for deploying and fine-tuning open-source models on AWS infrastructure. It directly addresses common production bottlenecks such as IAM permission configuration and GPU quota management, making it highly actionable for Dutch enterprises utilizing cloud-based AI.





