AI Project Life Cycle, MLOps, and DataOps CAIPM Practice Questions (Page 5)

Questions 21–25

1. 21expert · hard

   A healthcare company has a model that predicts patient readmission. The model is deployed and performs well, but the data distribution shifts over time due to changes in patient demographics. The ML team wants to implement continuous training, but the regulatory team requires that every model version be validated and documented before deployment. Which approach satisfies both requirements?

   • ADeploy every retrained model automatically without review
   • BAutomate the retraining pipeline but include a human approval step before deployment
   • CStop retraining and rely on the original model
   • DRetrain the model manually every quarter
   Reveal Solution Discussion Select an answer first
2. 22application · medium

   A healthcare startup is building a patient-readmission prediction model. The data engineering team uses an orchestration tool to schedule data pipelines, while the ML team uses a separate CI/CD pipeline to deploy the model. The two teams operate independently, and the model frequently fails in production because the data schema changed without the ML team being notified. Which approach best integrates DataOps and MLOps to prevent this issue?

   • AEstablish a shared pipeline that runs data validation and model deployment in the same workflow
   • BRequire the ML team to use the same orchestration tool as the data team
   • CAsk the data engineering team to freeze the data schema for the project
   • DHave the ML team manually check the data schema before each deployment
   Reveal Solution Discussion Select an answer first
4. 23foundation · easy

   Which tool category is commonly used in MLOps to manage changes to code, models, and data over time?

   • AVersion control systems
   • BMonitoring dashboards
   • CData orchestration platforms
   • DCI/CD pipelines
   Reveal Solution Discussion Select an answer first
5. 24foundation · easy

   Which scenario best illustrates the integration of MLOps and DataOps to improve operational efficiency?

   • ADataOps and MLOps operate independently with no shared processes
   • BMLOps builds the data pipeline, and DataOps trains the model
   • CDataOps manually uploads data files, and MLOps manually retrains models on a monthly basis
   • DDataOps automates data validation, and MLOps automatically triggers model retraining when new validated data is available
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   In DataOps, which principle is primarily concerned with fostering communication and shared responsibility among data engineers, data scientists, and business stakeholders?

   • ACollaboration
   • BMonitoring
   • COrchestration
   • DAutomation
   Reveal Solution Discussion Select an answer first