Plan & Manage
an Azure AI Solution
This is the foundation of the entire exam. Before you can build any AI service, you need to know how to choose the right one, set it up securely, keep it running smoothly, and make sure it behaves responsibly. Everything in Parts 2–6 depends on concepts introduced here.
Part 1 carries the highest single weight of any domain at 20–25%. Exam questions here are often scenario-based: "A company needs to process invoices while keeping data in a private network — which service and security configuration should they use?" You will not be able to answer these without knowing how Azure AI services are structured, secured, monitored, and governed. Additionally, terms like endpoint, API key, managed identity, and resource group appear in every single other domain — so mastering them here pays dividends across the entire exam.
What you'll learn in Part 1
Topics in this Part
Study these in order — each topic builds on the previous one.
Start here. Before writing a single line of code, you need to understand what Microsoft Foundry is, how all Azure AI services are organised under it, and — most importantly — how to choose the right service for a given scenario. This topic also introduces the foundational vocabulary (endpoint, API key, resource, subscription) that every other topic assumes you know.
Now that you know which service to pick, this topic covers how to actually create it in Azure. You'll learn how to provision a resource in the portal, choose the right AI model, understand deployment options (cloud, container, edge), and integrate your service into a CI/CD pipeline. All code examples are in Python with every line explained.
One of the most heavily tested areas. This topic covers every way to authenticate to an Azure AI service: API keys, Microsoft Entra ID tokens, managed identities (system-assigned vs user-assigned), and service principals. You'll also learn RBAC roles, Azure Key Vault integration, private endpoints, VNet configuration, and data encryption including customer-managed keys.
Once your service is running, you need to keep an eye on it. This topic covers Azure Monitor metrics (calls, errors, latency), Application Insights integration, diagnostic settings and log routing, writing KQL queries to analyse logs, setting up alert rules, and managing costs with budgets and pricing tiers.
Microsoft has strong opinions about responsible AI and the exam reflects that. This topic covers the 6 Responsible AI principles you must know by name, the Azure AI Content Safety service, content filters and blocklists, prompt shields (which detect jailbreak attempts), harm categories, and how to design a responsible AI governance framework.
You've chosen Python for your exam, which is the right choice — it has the best SDK support for Azure AI services. Every code example across all 18 topic files will be Python only, and every single line will have a comment explaining what it does. You do not need prior Python experience to follow along. Topic 1 will also introduce the two basic Python concepts you need before any code appears: importing a library and calling a function.