Summary

In this chapter, the focus was on deploying your model as a managed online endpoint to support real-time inferencing use cases in an automated fashion.

This chapter brought together the concepts that you learned about in previous chapters, as well as introducing you to Azure DevOps and the orchestration it makes possible. Leveraging Azure DevOps, code and deployments are traceable. Azure DevOps pipelines automate triggering the orchestration of the dev environment pipeline, moving the registered model to the higher environment. Leveraging Azure Key Vault, we can securely hold information to support multiple environments, linking those to your Azure DevOps environment groups. With MLflow integration, metrics are captured for the model generated in the dev environment, and that model number is registered in a higher environment and then added to the managed online endpoint. We implemented an MLOps pipeline to automate data transformation, model creation, evaluation, and model...