AI Project Life Cycle, MLOps, and DataOps CAIPM Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A logistics company has a demand-forecasting model that retrains weekly. The ML team wants to move to continuous training to capture rapid market changes, but the operations team is concerned about the cost of running training jobs more frequently and the risk of deploying untested models. Which approach best balances these concerns?

   • AImplement continuous training with automated validation gates and rollback capabilities
   • BKeep the weekly retraining schedule to avoid any risk
   • CDeploy every retrained model immediately to production
   • DUse a smaller training dataset to reduce cost and time
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A logistics company runs a data pipeline that ingests GPS data from delivery vehicles. The data engineering team wants to ensure that the pipeline is reliable and that any data quality issues are caught early. They decide to implement automated data quality checks. Which additional DataOps practice should they adopt to improve collaboration between data engineers and data scientists?

   • ARestrict access to the data pipeline to only data engineers
   • BHave data scientists manually inspect the data before each use
   • CEstablish a shared data catalog with documented schemas and quality metrics
   • DUse a different data pipeline for each team
   Reveal Solution Discussion Select an answer first
4. 18expert · hard

   A financial institution is implementing DataOps for its regulatory reporting pipelines. The pipelines must be highly reliable and auditable. The team is considering using a data orchestration tool. Which feature is MOST important for meeting regulatory requirements?

   • AIntegration with cloud storage services.
   • BThe ability to run pipelines on a schedule.
   • CDetailed logging and lineage tracking of data transformations.
   • DSupport for multiple programming languages.
   Reveal Solution Discussion Select an answer first
5. 19application · medium

   A retail company is building a real-time recommendation system. The data pipeline provides features to the model, and the model's predictions influence inventory decisions. The program manager wants to ensure that both data and model issues are detected quickly. Which approach BEST integrates MLOps and DataOps monitoring?

   • AMonitoring only model performance in production.
   • BMonitoring only data pipeline health.
   • CUnified monitoring that tracks data quality, feature drift, and model performance in one dashboard.
   • DSeparate monitoring for data and model with different teams.
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   Which role is primarily responsible for building and maintaining the data pipelines that feed machine learning models?

   • AData scientist
   • BOperations staff
   • CML engineer
   • DData engineer
   Reveal Solution Discussion Select an answer first