SageMaker Canvas: specification of the predicted results path in S3

Yes, you can specify the S3 path where Amazon SageMaker Canvas stores prediction results and other artifacts. This is configured during the setup of your SageMaker domain or user profile.

When setting up SageMaker Canvas, you have two options for storage configuration:

1. System managed: This uses the default SageMaker-created bucket that follows the pattern s3://sagemaker-{Region}-{your-account-id}. SageMaker creates a Canvas/ folder inside this bucket to store all Canvas data, including model artifacts, batch predictions, datasets, and logs.

2. Custom S3: You can specify your own Amazon S3 bucket as the storage location by entering a custom Amazon S3 URI. This gives you more control over where your Canvas data is stored.

You can configure this storage location at two levels:

• At the domain level during domain setup (applies to all users)
• At the user profile level (overrides the domain-level configuration)

Additionally, you can optionally specify a KMS key for encrypting Canvas artifacts stored at the specified location.

For asynchronous inference specifically, when creating an endpoint configuration, you can use the AsyncInferenceConfig object to specify an S3 output path where you want the model's output to be saved using the S3OutputPath parameter.

Note that if you update the storage location for an existing Canvas application, you'll lose access to your previous artifacts from the old S3 location within the Canvas interface, though you can still access them directly in Amazon S3.
Sources
Getting started with using Amazon SageMaker Canvas - Amazon SageMaker AI
Configure your Amazon S3 storage - Amazon SageMaker AI
How do you pass an S3 path as a parameter in an asynchronous inference call to a model in SageMaker? | AWS re:Post