AI Azure AI Engineer 2 — Questions and Answers
Question 1: Which Azure service provides a managed environment for training and deploying machine learning models at scale?
- Azure Databricks
- Azure Machine Learning (Correct answer)
- Azure Synapse Analytics
- Azure HDInsight
Correct answer: Azure Machine Learning
Azure Machine Learning is the fully managed platform for the end-to-end ML lifecycle including training, deployment, and monitoring.
Question 2: In Azure Cognitive Services, which API would you use to extract key phrases and sentiment from customer reviews?
- Computer Vision API
- Text Analytics API (Correct answer)
- Translator API
- Speech API
Correct answer: Text Analytics API
Text Analytics API (now part of Azure AI Language) provides sentiment analysis, key phrase extraction, and named entity recognition for text.
Question 3: What is the purpose of an Azure Machine Learning compute cluster compared to a compute instance?
- Compute clusters are for interactive notebooks; compute instances are for batch jobs
- Compute clusters scale to zero when idle and support parallel jobs; compute instances are single-node development VMs (Correct answer)
- Compute clusters run inference; compute instances run training
- Compute clusters are serverless; compute instances require manual provisioning
Correct answer: Compute clusters scale to zero when idle and support parallel jobs; compute instances are single-node development VMs
Compute clusters auto-scale nodes up and down for batch/parallel training jobs, while compute instances are always-on single-node VMs for development.
Question 4: Which deployment option in Azure AI Foundry allows you to host an open-source model like Llama 3 on dedicated infrastructure?
- Serverless API endpoint
- Managed compute endpoint (Correct answer)
- Azure OpenAI deployment
- Azure Functions
Correct answer: Managed compute endpoint
Managed compute endpoints provision dedicated GPU infrastructure to host open-source models, giving you full control over the runtime.
Question 5: When building a RAG (Retrieval-Augmented Generation) solution on Azure, which service is typically used to store and search document embeddings?
- Azure Cosmos DB for NoSQL
- Azure AI Search (Correct answer)
- Azure Table Storage
- Azure Cache for Redis
Correct answer: Azure AI Search
Azure AI Search (formerly Cognitive Search) provides vector search capabilities that enable semantic retrieval of document chunks for RAG pipelines.
Question 6: An Azure AI Engineer needs to prevent a language model from generating harmful content. Which Azure OpenAI feature addresses this directly?
- Prompt caching
- Content filters (Correct answer)
- Token limits
- Temperature settings
Correct answer: Content filters
Azure OpenAI content filters use multi-category classifiers to detect and block harmful content in both prompts and completions.
Question 7: Which metric best measures the quality of an object detection model's ability to correctly locate and classify objects?
- Accuracy
- Mean Average Precision (mAP) (Correct answer)
- F1 Score
- AUC-ROC
Correct answer: Mean Average Precision (mAP)
mAP accounts for both localization (bounding box overlap) and classification correctness across multiple IoU thresholds and object classes.
AI Azure AI Engineer 2
Which Azure service provides a managed environment for training and deploying machine learning models at scale?