Draft AI Solution PMI-CPMAI Practice Questions (Page 3)

Questions 11–15

1. 11foundation · easy

   In a high-level AI architecture diagram, which component is typically responsible for receiving raw input data from external systems and making it available for downstream processing?

   • AModel serving layer
   • BUser interface layer
   • CFeature engineering layer
   • DData ingestion layer
   Reveal Solution Discussion Select an answer first
2. 12application · medium

   A government agency wants to deploy an AI system for document classification. The data is highly sensitive and must not leave the agency's premises. The system will be used by employees via a web interface. Which deployment strategy and architecture component are most appropriate?

   • APublic cloud deployment with a virtual private cloud (VPC) and strict access controls.
   • BOn-premises deployment with a private cloud environment, including a web application server and a model serving API.
   • CEdge deployment on individual employee laptops, with the model running locally.
   • DHybrid deployment with data stored on-premises but model training in the cloud.
   Reveal Solution Discussion Select an answer first
4. 13foundation · easy

   In the data flow pipeline, which stage is primarily responsible for transforming raw data into a format suitable for model training, such as creating numerical representations from text?

   • AModel deployment
   • BFeature engineering
   • CData storage
   • DData ingestion
   Reveal Solution Discussion Select an answer first
5. 14application · medium

   A healthcare provider wants to identify patients at high risk of readmission within 30 days of discharge. They have electronic health records (EHR) with labeled outcomes from past admissions. The solution must be explainable to clinicians and integrate with the EHR system. Which AI model type and architecture component are most appropriate?

   • AReinforcement learning to optimize treatment plans, with a streaming pipeline that ingests live patient vitals and serves recommendations via a REST API.
   • BSupervised classification using a deep neural network, with a batch pipeline that ingests EHR data and serves predictions via a REST API, but without model explainability features.
   • CSupervised classification using logistic regression, with a batch pipeline that ingests EHR data, trains the model, and serves predictions via a REST API to the EHR system.
   • DUnsupervised clustering to group patients, with a real-time pipeline that ingests EHR data and serves cluster labels via a REST API.
   Reveal Solution Discussion Select an answer first
6. 15foundation · easy

   When documenting integration points with an existing system, which element is essential to specify for the data exchange?

   • AThe brand of the server hardware
   • BThe data exchange format and protocol
   • CThe number of developers on the team
   • DThe color scheme of the user interface
   Reveal Solution Discussion Select an answer first