AI Azure AI Engineer 5 — Questions and Answers
Question 1: A Responsible AI principle requires that AI system decisions can be understood and traced. Which principle does this describe?
- Fairness
- Reliability
- Transparency (Correct answer)
- Inclusiveness
Correct answer: Transparency
Transparency means AI systems and their decision-making processes should be understandable to those who design, deploy, and use them.
Question 2: Which Azure resource provides a centralized hub for managing connections, compute, and deployments across Azure AI services within a project?
- Azure Resource Group
- Azure AI Foundry Hub (Correct answer)
- Azure Management Group
- Azure Subscription
Correct answer: Azure AI Foundry Hub
Azure AI Foundry Hub is the top-level organizational resource that provides shared compute, connections, and governance for AI projects within an organization.
Question 3: An engineer needs to redact personally identifiable information (PII) from documents before storing them. Which Azure AI service provides this capability natively?
- Azure Purview
- Azure AI Language with PII detection (Correct answer)
- Azure Key Vault
- Azure Information Protection
Correct answer: Azure AI Language with PII detection
Azure AI Language includes a PII detection and redaction feature that identifies and masks entities like names, SSNs, and phone numbers in text.
Question 4: What does the 'grounding' technique in prompt engineering accomplish when working with Azure OpenAI?
- Reduces the model's hallucination by injecting relevant source documents into the prompt context (Correct answer)
- Forces the model to output only factual Wikipedia entries
- Limits the model to a specific token count
- Prevents the model from using training data older than a certain date
Correct answer: Reduces the model's hallucination by injecting relevant source documents into the prompt context
Grounding provides the model with retrieved context from authoritative sources, anchoring its responses to verified information and reducing hallucination.
Question 5: In Azure Custom Vision, what is the minimum recommended number of images per tag for training an image classification model?
- 5
- 15 (Correct answer)
- 50
- 200
Correct answer: 15
Azure Custom Vision recommends at least 15 images per tag to train a meaningful classifier, though more diverse images improve accuracy significantly.
Question 6: Which Azure networking feature ensures that traffic between your Azure ML workspace and storage accounts does not traverse the public internet?
- Azure DDoS Protection
- Private endpoints with Virtual Network integration (Correct answer)
- Azure Front Door
- Network Security Groups only
Correct answer: Private endpoints with Virtual Network integration
Private endpoints assign a private IP to Azure services within your VNet, routing traffic over the Microsoft backbone rather than the public internet.
Question 7: When using Azure AI Vision's Image Analysis 4.0 API, which capability allows you to ask natural language questions about the content of an image?
- OCR Read API
- Dense Captioning
- Visual question answering (VQA) (Correct answer)
- Smart Crop
Correct answer: Visual question answering (VQA)
Visual question answering lets you pass an image along with a natural language question and receive a contextually relevant answer from the model.
AI Azure AI Engineer 5
A Responsible AI principle requires that AI system decisions can be understood and traced.
Which principle does this describe?