Azure-AI_Foundry

Module 1: Knowledge Check

Test your understanding of Azure AI Foundry fundamentals. Choose the best answer for each question.

Question 1

Your company is starting a new generative AI initiative with multiple teams working on different applications. All teams need access to GPT-4o and Azure AI Search. What is the recommended resource structure?

A) Create a separate AI Hub for each team
B) Create one AI Hub with shared connections and a separate Project for each team
C) Create one Project and share it across all teams
D) Skip AI Foundry and use standalone Azure OpenAI resources

✅ Answer **B) Create one AI Hub with shared connections and a separate Project for each team** Hubs provide shared infrastructure and connections, while projects isolate team workspaces. This maximizes resource sharing while maintaining separation of concerns.

Question 2

Which underlying Azure resource does AI Foundry use to securely store connection strings and API keys?

A) Azure Storage Account
B) Azure Container Registry
C) Azure Key Vault
D) Azure App Configuration

✅ Answer **C) Azure Key Vault** Key Vault is automatically provisioned with each hub and stores all secrets, connection strings, and API keys used by connections.

Question 3

You need to deploy an AI Foundry hub that is only accessible from your corporate network. Which networking option should you choose?

A) Public access with IP allowlist
B) Managed Virtual Network with private endpoints
C) Azure Front Door
D) Network Security Group only

✅ Answer **B) Managed Virtual Network with private endpoints** Managed VNet with private endpoints ensures the hub is only accessible from your corporate network via Azure Private Link, providing enterprise-grade network isolation.

Question 4

What is the purpose of a Connection in Azure AI Foundry?

A) To link two AI Hubs together
B) To connect a Project to an external Azure or third-party resource
C) To establish a VPN tunnel to on-premises systems
D) To synchronize data between regions

✅ Answer **B) To connect a Project to an external Azure or third-party resource** Connections securely link external resources (Azure OpenAI, AI Search, Storage, custom APIs) to the hub, making them available to all child projects.

Question 5

Which Python package should you install to make chat completion calls to models deployed in AI Foundry?

A) azure-ai-ml
B) azure-ai-projects
C) azure-ai-inference
D) azure-cognitiveservices-language

✅ Answer **C) `azure-ai-inference`** The `azure-ai-inference` package provides the client for calling deployed models (chat completions, embeddings, image generation). `azure-ai-ml` is for workspace management, and `azure-ai-projects` is for project-level operations.

Question 6

You are deploying infrastructure for AI Foundry using Bicep. What kind property do you set on the Microsoft.MachineLearningServices/workspaces resource to create a hub?

A) kind: 'Workspace'
B) kind: 'Hub'
C) kind: 'AIServices'
D) kind: 'Project'

✅ Answer **B) `kind: 'Hub'`** In Bicep, the `Microsoft.MachineLearningServices/workspaces` resource type uses `kind: 'Hub'` for hubs and `kind: 'Project'` for projects.

Question 7

What is the recommended authentication approach when developing with the Azure AI Foundry SDK?

A) Embed API keys directly in source code
B) Use DefaultAzureCredential from azure-identity
C) Store keys in environment variables and read them at runtime
D) Use shared access signatures (SAS tokens)

✅ Answer **B) Use `DefaultAzureCredential` from `azure-identity`** `DefaultAzureCredential` supports managed identity in production and developer credentials locally, eliminating the need for API keys in code. This is the recommended approach for all Azure SDK interactions.

Question 8

A developer wants to quickly test GPT-4o with different system prompts before writing any application code. Which AI Foundry feature should they use?

A) Prompt Flow
B) Model Catalog
C) Chat Playground
D) Evaluation

✅ Answer **C) Chat Playground** The Chat Playground in the AI Foundry portal allows interactive testing of models with different system prompts, parameters, and configurations — ideal for prototyping before writing code.

Question 9

Your organization has strict compliance requirements that mandate all AI resources in Europe must be separate from those in North America. How should you structure your AI Foundry resources?

A) One hub with projects in different regions
B) Separate hubs in each region
C) One project with multi-region deployments
D) Use Azure Front Door to route between regions

✅ Answer **B) Separate hubs in each region** Since hubs are the governance boundary, creating separate hubs in each region ensures complete isolation of infrastructure, data, and security policies to meet compliance requirements.

Question 10

Which of the following is NOT automatically created when you provision an AI Foundry Hub?

A) Storage Account
B) Key Vault
C) Azure SQL Database
D) AI Services resource

✅ Answer **C) Azure SQL Database** An AI Foundry Hub automatically creates a Storage Account, Key Vault, and AI Services resource. Container Registry and Application Insights are optional. Azure SQL Database is not part of the default provisioning.

Question 11

You want to use both GPT-4o from OpenAI and Llama 3.1 from Meta in the same application. What feature of AI Foundry makes this possible?

A) Model fine-tuning
B) Model catalog with multiple deployment types
C) Prompt Flow only
D) Custom containers

✅ Answer **B) Model catalog with multiple deployment types** The model catalog supports models from multiple providers. GPT-4o deploys as an Azure OpenAI resource, while Llama 3.1 can deploy as a serverless API (Model as a Service), both accessible from the same project.

Question 12

What is the relationship between an AI Hub and an AI Project?

A) They are independent resources with no relationship
B) A Project contains multiple Hubs
C) A Hub contains multiple Projects and provides shared infrastructure
D) They are the same resource with different views

✅ Answer **C) A Hub contains multiple Projects and provides shared infrastructure** The Hub is the parent resource providing shared connections, security, and infrastructure. Projects are child workspaces where teams build and deploy their specific AI solutions.

Question 13

Which CLI extension must you install to manage AI Foundry resources from the Azure CLI?

A) az extension add --name ai
B) az extension add --name ml
C) az extension add --name cognitiveservices
D) az extension add --name ai-foundry

✅ Answer **B) `az extension add --name ml`** AI Foundry uses the Azure Machine Learning control plane, so the `ml` CLI extension is required for hub and project management operations.

Question 14

A team is building a RAG-based chatbot that needs Azure OpenAI for LLM calls and Azure AI Search for document retrieval. Where should these service references be configured?

A) As environment variables in the application
B) As Connections at the Hub level
C) As separate resource groups
D) As Kubernetes secrets

✅ Answer **B) As Connections at the Hub level** Hub-level connections securely store credentials in Key Vault and make the linked services available to all child projects. This centralizes configuration and simplifies access management.

Question 15

You deployed an AI Foundry hub in East US. Your team in London reports high latency when using the playground. What is the recommended solution?

A) Upgrade to a higher SKU
B) Create an additional hub in a European region (e.g., UK South or Sweden Central)
C) Enable CDN for the hub
D) Increase the rate limit on the deployment

✅ Answer **B) Create an additional hub in a European region (e.g., UK South or Sweden Central)** Each hub is region-specific. For global teams, deploy hubs in regions closest to your users to minimize latency. Model availability varies by region, so check the availability matrix.

Score Guide

Score Rating
13–15 ⭐ Excellent — Ready for Module 2
10–12 👍 Good — Review missed topics
7–9 📖 Fair — Re-read lessons before continuing
< 7 🔄 Needs review — Revisit all lessons
*← Lesson 2 Module 2 — Projects & Hubs →*