MLOps Best Practices MACHINE-LEARNING-ASSOCIATE Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A logistics company is starting an MLOps initiative to manage its delivery time prediction models. The team has limited engineering resources and wants to start small while still gaining benefits. What is a reasonable first step?

   • AHire a dedicated MLOps engineer before any automation is in place.
   • BWait until the team has more models before starting MLOps.
   • CImplement automated model retraining and deployment for one pilot model.
   • DBuild a full CI/CD pipeline for all models at once.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A technology company is defining its MLOps strategy for a new product. The team is small and has limited experience with MLOps. They want to adopt best practices but are concerned about the learning curve and time to implementation. What is the most pragmatic approach?

   • AStart with a simple version of MLOps, such as using MLflow for tracking and the model registry, and gradually add automation.
   • BAvoid MLOps tools and rely on manual processes to save time.
   • CImplement a full CI/CD pipeline with automated testing and deployment from the start.
   • DHire a large MLOps team before starting any implementation.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   What is a key benefit of defining an MLOps strategy for an organization?

   • AIt guarantees that models will never fail in production
   • BIt eliminates the need for data scientists to collaborate with engineers
   • CIt provides a consistent process for moving models from experimentation to production
   • DIt ensures that all models are trained using the same dataset
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A retail company is deploying a demand forecasting model. The data science team wants to ensure that the model can be retrained and deployed with minimal manual intervention. Which practice is most aligned with MLOps best practices?

   • AHave a data scientist manually deploy the model after each training run.
   • BUse a shared notebook for training and deployment without version control.
   • CCreate a CI/CD pipeline that automatically builds, tests, and deploys the model.
   • DRetrain the model only when a business stakeholder requests it.
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A healthcare startup is deploying a patient readmission prediction model. The compliance team requires that every model prediction be traceable to the exact model version and input data. The data science team uses MLflow for experiment tracking and the model registry. What is the most effective way to meet this requirement?

   • ALog the model version and input data hash in the serving infrastructure's request logs.
   • BStore the model in the registry but do not log predictions.
   • CUse a single model version for all predictions to simplify tracking.
   • DRely on the data scientist's memory of which model was used.
   Reveal Solution Discussion Select an answer first