Deploying hubs and projects with Microsoft Foundry
5 minutes
5 Questions
Microsoft Foundry provides a comprehensive platform for deploying AI hubs and projects within the Azure ecosystem. Azure AI Foundry serves as the central workspace where teams can build, deploy, and manage generative AI solutions effectively.
An AI Hub acts as a top-level resource that provides sh…Microsoft Foundry provides a comprehensive platform for deploying AI hubs and projects within the Azure ecosystem. Azure AI Foundry serves as the central workspace where teams can build, deploy, and manage generative AI solutions effectively.
An AI Hub acts as a top-level resource that provides shared infrastructure, security settings, and governance for multiple AI projects. When deploying a hub, you configure essential elements including the Azure subscription, resource group, region, and networking settings. Hubs enable centralized management of connections to Azure services like Azure OpenAI, Azure AI Search, and storage accounts.
Projects exist within hubs and represent individual AI applications or workloads. Each project inherits security and connection configurations from its parent hub while maintaining isolation for specific development activities. When creating a project, you specify the project name, description, and associated hub.
The deployment process through Azure AI Foundry portal involves several steps. First, navigate to the Azure AI Foundry portal and select Create new hub. Configure the hub settings including name, subscription, resource group, and region. Enable managed identity for secure authentication. Next, create projects within the hub by selecting New project and providing project details.
For programmatic deployment, you can use Azure CLI, PowerShell, or Infrastructure as Code tools like Bicep and ARM templates. The Azure CLI command az ml workspace create with appropriate parameters enables hub creation, while similar commands handle project provisioning.
Key considerations during deployment include selecting appropriate compute resources, configuring private endpoints for network security, setting up role-based access control for team members, and establishing connections to required Azure services. Proper planning ensures scalability, security, and cost optimization.
After deployment, teams can leverage the hub and projects to develop prompt flows, fine-tune models, conduct evaluations, and deploy generative AI applications to production endpoints with built-in monitoring capabilities.
Deploying Hubs and Projects with Microsoft Foundry
Why This Topic Is Important
Understanding how to deploy hubs and projects in Azure AI Foundry is essential for the AI-102 exam because it represents the foundational infrastructure for building generative AI solutions. Microsoft has positioned Azure AI Foundry as the primary platform for enterprise AI development, making this knowledge critical for real-world implementations and exam success.
What Are Hubs and Projects in Azure AI Foundry?
Azure AI Hub is a top-level resource that provides a centralized location for managing AI resources, security settings, and shared configurations across multiple projects. Think of it as a container that houses common settings and resources.
Azure AI Projects are workspaces within a hub where actual development occurs. Each project inherits settings from its parent hub while allowing for project-specific configurations.
Key Components: • Hub: Manages shared resources like compute, connections, and security policies • Project: Contains models, deployments, prompt flows, and evaluation assets • Connections: Store credentials for Azure OpenAI, Azure AI Search, and other services • Compute: Resources for running models and prompt flows
How Deployment Works
Step 1: Create an Azure AI Hub Navigate to the Azure portal and create a new Azure AI Hub resource. Configure the region, resource group, and associated storage account.
Step 2: Configure Hub Settings Set up shared connections to Azure OpenAI Service, Azure AI Search, and other required services. Define security policies and network configurations.
Step 3: Create Projects Within the Hub Create one or more projects under the hub. Each project can have its own deployments while sharing hub-level resources.
Step 4: Deploy Models Within projects, deploy foundation models from the model catalog or custom fine-tuned models.
• One hub can contain multiple projects • Projects inherit hub connections and policies • Resources like Azure OpenAI connections are defined at the hub level • Model deployments and prompt flows exist at the project level
Exam Tips: Answering Questions on Deploying Hubs and Projects
Tip 1: Remember the hierarchy - Hubs are parent resources that contain Projects. If a question asks about shared resources or centralized management, the answer typically involves the Hub.
Tip 2: Questions about security policies, network isolation, or managed identities usually relate to Hub-level configurations.
Tip 3: When asked about where model deployments or prompt flows reside, the answer is at the Project level.
Tip 4: Pay attention to scenarios requiring resource sharing across teams - this indicates Hub-level resource management.
Tip 5: For infrastructure-as-code questions, know that both ARM templates and Bicep support Azure AI Foundry resource deployment.
Tip 6: Understand that connections (credentials to external services) are typically created at the Hub level and inherited by Projects.
Tip 7: If a question mentions cost management or resource governance across multiple AI projects, think Hub-level policies.
Common Exam Scenarios: • Choosing between creating a new hub vs. a new project • Selecting appropriate deployment methods for different requirements • Understanding which resources belong at hub vs. project level • Configuring connections and security settings
Key Terms to Remember: • Azure AI Foundry portal: The web interface for managing hubs and projects • Model catalog: Repository of available foundation models • Managed compute: Azure-managed resources for running workloads • Connections: Stored credentials for integrated services