How to cancel SageMaker

I understand your concern about the unexpected SageMaker billing and the high forecast for the current month. To address this issue and cancel SageMaker resources, you'll need to take several steps:

1. Identify and stop running resources:

   • Use AWS Cost Explorer to pinpoint which specific SageMaker resources are causing the high costs.
   • Stop or delete any running notebook instances, Studio instances, inference endpoints, training jobs, or experiments.

2. Clean up hidden resources:

   • Check for resources like SageMaker Canvas workspaces that might be running in the background.
   • Ensure all apps associated with user profiles are stopped or deleted.

3. Delete SageMaker Domain and user profiles:

   • After stopping all running resources, delete any user profiles.
   • Once user profiles are removed, delete the SageMaker Domain.

4. Cancel any subscriptions:

   • If you have any ML algorithm or model package subscriptions, cancel them in the AWS Marketplace.
   • Before canceling, terminate any running training jobs, real-time endpoints, or batch inference jobs associated with these subscriptions.

5. Double-check all SageMaker features:

   • Review and shut down resources for all SageMaker AI features, including Canvas, Code Editor, EMR in Studio, HyperPod, Pipelines, and RStudio.

6. Set up billing alerts:

   • Configure AWS Budgets to alert you before exceeding your desired spending limit in the future.

If you're still having trouble identifying or deleting specific resources, it's advisable to contact AWS Support. They can provide a detailed breakdown of your charges and help identify any lingering resources that may be causing the high billing forecast.

Remember, for future use, always monitor your usage closely and shut down resources promptly after use to avoid unexpected charges.
Sources
Sagemaker has pushed us over budget. | AWS re:Post
Where to shut down resources per SageMaker AI features - Amazon SageMaker AI
Canceling product subscriptions - AWS Marketplace