AI-102 Question 48
Select 3You are an AI engineer tasked with integrating an Azure Cognitive Services Text Analytics model into your company's CI/CD pipeline. The pipeline must ensure that the model is retrained with updated data, validated for accuracy, and deployed automatically to the production environment. Which steps should you include in your CI/CD pipeline to achieve this?
- A
Configure a training pipeline in Azure Machine Learning to retrain the model using updated data.
- B
Use an Azure DevOps release pipeline to deploy the updated model to production.
- C
Manually trigger the deployment of the model to production after retraining it.
- D
Set up automated validation to test the model's accuracy before production deployment.
- E
Create a custom web app to manually monitor the model's performance and deploy it as needed.
Show answer and explanation
Correct answers: A, B, D
Explanation
Integrating Azure AI services into a CI/CD pipeline involves automating the entire workflow, from retraining the model with updated data to deploying it into production. Using Azure Machine Learning for training, Azure DevOps pipelines for deployment, and automated validation ensures a seamless and scalable process. Manual steps, such as triggering deployments or monitoring, are not aligned with the goals of CI/CD automation.
- A. Correct.
Correct. Setting up a training pipeline in Azure Machine Learning ensures that the model is retrained with updated data as part of the CI/CD process.
- B. Correct.
Correct. An Azure DevOps release pipeline automates the deployment of the updated model to the production environment, making it part of the CI/CD workflow.
- C. Incorrect.
Incorrect. Manually triggering deployments breaks the automation principle of CI/CD and is not recommended in a robust pipeline.
- D. Correct.
Correct. Automating the validation step ensures that only models meeting accuracy requirements are deployed to production, aligning with CI/CD best practices.
- E. Incorrect.
Incorrect. While monitoring is important, manually deploying the model through a custom app does not align with CI/CD automation principles.