Hugging Face to SageMaker: One-click model deployment and fine-tuning

Amazon Web Services and Hugging Face have unveiled a deep-link integration that collapses the multi-step process of moving from model discovery to enterprise deployment into a single click. Now, when browsing supported models on Hugging Face, developers see two new action buttons: "Customize on SageMaker AI" and "Deploy on SageMaker AI." Clicking either sends them directly into SageMaker Studio with the selected model pre-loaded, permissions pre-configured, and the environment fully ready for experimentation. The new managed policy, AmazonSageMakerModelCustomizationCoreAccess, automatically grants permissions for serverless fine-tuning jobs using supervised fine-tuning (SFT), direct preference optimization (DPO), reinforcement learning with verifiable rewards (RLVR), and reinforcement learning from AI feedback (RLAIF). Additionally, GPU quota for instance types like G5 and G6 is now surfaced directly in the Studio UI, eliminating the need to navigate to Service Quotas separately.

Previously, developers had to open the AWS Console, create a SageMaker domain, configure IAM roles and policies, and sometimes request GPU quota increases before they could even begin experimenting. This friction is now removed, significantly accelerating the path from inspiration to iteration. Mark McQuade, CEO of Arcee AI, praised the integration: "Going from an open model on Hugging Face straight into SageMaker Studio in a single click, then fine-tuning or deploying it inside your own AWS environment with nothing to wire up, is the kind of experience open models have been missing." For enterprises, this means a faster, more secure way to customize and deploy open-weight models while retaining control over data and infrastructure. The integration is available now for supported models on Hugging Face.