Building the state machine workflow
From the Deploying the CI/CD pipeline section of Chapter 6, Automating the Machine Learning Process Using AWS Step Functions, you will recall that we deployed a CI/CD pipeline to orchestrate the E2E ML process as a Cloud Development Kit (CDK) application. However, as you saw in Figure 6.3, the abalone-cicd-pipeline execution failed as there were no ModelSource artifacts.
Consequently, it's at this stage of the overall process that the ML practitioner must create these source artifacts to build the ML workflow, using the AWS Step Functions Data Science SDK for Python. We will therefore switch our perspective to that of the ML practitioner and build these source artifacts, using the SageMaker Studio user interface (UI).
Setting up the service permissions
Before we can begin defining the state machine workflow within a Jupyter Notebook, we need to assign additional access permissions for the SageMaker execution role to accommodate the...
