AI Platforms, Tools, and Ecosystem Integration CAIPM Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   An online media company runs a real-time content recommendation model that experiences latency spikes during peak traffic. The model is deployed on a fixed-size cluster. The program manager needs to improve performance without over-provisioning during off-peak hours. Which strategy should they implement?

   • AEnable auto-scaling on the cluster based on CPU utilization and request latency
   • BIncrease the cluster size permanently to handle peak traffic
   • CReduce the model's input features to lower latency
   • DMove the model to a batch processing system that runs every hour
   Reveal Solution Discussion Select an answer first
2. 7application · medium

   A financial services firm is building a data pipeline to feed a fraud detection model. The pipeline must ingest real-time transaction data, combine it with historical customer profiles, and deliver features to the model with low latency. Which pipeline design best meets these requirements?

   • AA batch pipeline that runs nightly to update a feature store used by the model
   • BA pipeline that sends raw transaction data directly to the model without any feature engineering
   • CA streaming pipeline that processes events in real time and joins them with pre-computed profile features from a low-latency store
   • DA pipeline that stores all data in a data lake and runs ad-hoc queries for each prediction
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   Which stage in a data pipeline typically involves cleaning and transforming raw data into a usable format?

   • AModel monitoring
   • BModel deployment
   • CData processing
   • DData ingestion
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   Which component acts as an intermediary to translate and route requests between an application and multiple AI services?

   • ATraining cluster
   • BModel registry
   • CData warehouse
   • DAPI gateway
   Reveal Solution Discussion Select an answer first
6. 10application · medium

   A healthcare provider wants to integrate a third-party natural language processing (NLP) service to transcribe clinical notes. The integration must ensure that protected health information (PHI) is not exposed to the vendor's model training. Which approach should be used?

   • AUse a middleware layer that de-identifies PHI before sending data to the NLP service
   • BUse a direct API call from the EHR system to the NLP service without any intermediary
   • CSend the full clinical notes to the NLP service and rely on the vendor's privacy policy
   • DStore the clinical notes in the vendor's cloud for faster processing
   Reveal Solution Discussion Select an answer first