0

I have tried using the AWS::SageMaker::Pipeline resource in CloudFormation.

I want to give a pipeline definition in JSON format in CloudFormation. But there aren't any documentation available for that. There is documentation only for a Python SDK pipeline definition.

How can I create an MLOps SageMaker pipeline using CloudFormation?

Peter Mortensen
  • 30,738
  • 21
  • 105
  • 131
Harita
  • 16
  • 1
  • the easiest way would be to create a pipeline using the Python SDK, then saving it as JSON, using the `definition()` function. https://sagemaker.readthedocs.io/en/stable/workflows/pipelines/sagemaker.workflow.pipelines.html#pipeline . Then either use that, or at least see how it works if you want to DIY. It's a complicated structure though... – Neil McGuigan Feb 23 '22 at 18:07
  • Is the MLOps part relevant to the question (not a rhetorical question)? – Peter Mortensen Mar 25 '22 at 16:35

1 Answers1

1

This process is documented here.

You can either construct the pipeline definition using the SageMaker Python SDK or by writing the JSON definition directly. The SDK makes it easier to define a pipeline and get the JSON definition.

Once you have the definition in JSON, you can define the pipeline in CloudFormation as it is documented in the link above.

Peter Mortensen
  • 30,738
  • 21
  • 105
  • 131
Kirit Thadaka
  • 429
  • 2
  • 5