AI Platforms, Tools, and Ecosystem Integration CAIPM Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A streaming service has a recommendation model that is deployed on a cluster. The model's latency increases during peak viewing hours, and the operations team wants to reduce latency without increasing costs significantly. The team has identified that the model's inference time is the bottleneck, not the data loading. What should the team do?

   • ACache the model's predictions for common requests to reduce the number of inference calls
   • BIncrease the number of replicas of the model to handle more concurrent requests
   • CMove the model to a larger instance type to provide more CPU and memory
   • DOptimize the model by using quantization or pruning to reduce inference time, and enable auto-scaling for the cluster
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   Which of the following is a key component of an AI ecosystem that provides the raw material for training models?

   • ADeployment infrastructure
   • BOrchestration tools
   • CData sources
   • DMonitoring dashboards
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   When integrating an AI platform with an existing enterprise system, which approach is most appropriate for ensuring that data flows seamlessly between the two?

   • AUsing manual data entry
   • BDisabling all APIs
   • CReplacing the enterprise system
   • DUsing middleware and connectors
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A large insurance company has a production AI model that processes claims. The model is updated quarterly, and the company has a strict requirement to maintain an audit trail of all model versions and their performance. The data science team wants to test a new model version on a small percentage of live traffic before full deployment. The company also needs to ensure that the model's decisions can be explained to regulators. What should the company implement?

   • AA model registry with version tracking, a canary deployment strategy, and integrated explainability tools
   • BA shadow deployment where the new model runs in parallel but its decisions are not used, with no version tracking
   • CA manual approval process for each model update, with logs stored in a spreadsheet
   • DA simple deployment process that overwrites the old model with the new one and relies on the vendor's documentation for explainability
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   In an AI ecosystem, which component is responsible for executing the trained model to generate predictions in production?

   • ADeployment infrastructure
   • BModels
   • CDevelopment tools
   • DData sources
   Reveal Solution Discussion Select an answer first