AI-102 Question 47
Select 4You are developing a machine learning model using Azure Machine Learning and want to integrate it into a CI/CD pipeline for automated deployment. The pipeline should ensure that the model is registered in the Azure Machine Learning workspace, tested, and deployed to an Azure Kubernetes Service (AKS) cluster. Which of the following steps should you include in the pipeline?
- A
Use the Azure CLI to register the model in the Azure Machine Learning workspace.
- B
Configure a release pipeline task to deploy the model to an AKS cluster using an Azure Machine Learning endpoint.
- C
Manually test the model locally before deploying to the CI/CD pipeline.
- D
Integrate an automated model evaluation step to validate the model's performance before deployment.
- E
Use Azure DevOps to create a build pipeline that packages the model and its dependencies into a Docker container.
Show answer and explanation
Correct answers: A, B, D, E
Explanation
Integrating Azure AI services into a CI/CD pipeline requires automated processes for model registration, evaluation, packaging, and deployment. Steps like automated model evaluation and containerization ensure the pipeline adheres to DevOps principles. Manual testing is not suitable for CI/CD as it does not provide consistency or scalability.
- A. Correct.
Correct. Registering the model in the Azure Machine Learning workspace is essential to track and manage the model versions.
- B. Correct.
Correct. Configuring a release pipeline to deploy the model to an AKS cluster ensures the model is made available in a scalable environment for production use.
- C. Incorrect.
Incorrect. Manually testing the model is not a scalable solution and does not align with CI/CD practices, which emphasize automation.
- D. Correct.
Correct. Automated model evaluation ensures that only models meeting performance criteria are deployed, maintaining the quality of deployments.
- E. Correct.
Correct. Packaging the model and its dependencies into a Docker container prepares it for deployment in environments like AKS, aligning with DevOps practices.