Microsoft Azure AI Fundamentals (AI-900): 5 Essential Domains

The Azure AI Fundamentals exam (AI-900) validates foundational knowledge of AI and machine learning concepts plus Microsoft Azure AI services. These five domains map directly to the exam outline — use them as your study spine.

1. AI Workloads & Responsible AI Principles#

Understanding what AI can do—and how to do it responsibly—is foundational. This domain covers common AI workloads and Microsoft's framework for responsible AI.

Common AI Workloads#

Responsible AI Principles#

Microsoft's six principles for responsible AI development:

1. Fairness — AI systems should treat all people equitably
2. Reliability & Safety — AI should perform reliably and safely
3. Privacy & Security — AI should be secure and respect privacy
4. Inclusiveness — AI should empower everyone and engage people
5. Transparency — AI systems should be understandable
6. Accountability — People should be accountable for AI systems

2. Machine Learning Fundamentals#

Machine learning is the engine behind most AI systems. The exam tests your understanding of ML types, core concepts, and Azure Machine Learning capabilities.

Types of Machine Learning#

Supervised Learning: Classification vs Regression#

Unsupervised Learning: Clustering#

Clustering groups similar data points together without predefined labels:

• K-Means — Groups data into K clusters
• Hierarchical clustering — Creates a tree of clusters
• DBSCAN — Finds clusters of varying density

Azure Machine Learning Capabilities#

Azure Machine Learning provides a complete platform for the ML lifecycle:

• Designer — Drag-and-drop ML pipeline builder (no code)
• Automated ML (AutoML) — Automatically trains and tunes models
• Notebooks — Jupyter notebooks for custom code
• MLflow — Track experiments and manage model lifecycle
• Managed Endpoints — Deploy models for real-time or batch inference

3. Computer Vision on Azure#

Computer vision enables machines to interpret visual information. Azure provides both pre-built and customizable vision capabilities.

Computer Vision Task Types#

Azure AI Vision Services#

Custom Vision: Classification vs Object Detection#

• Classification — Predicts what's in the entire image ("This is a defective product")
• Object Detection — Finds multiple objects and their locations ("3 scratches at these coordinates")

4. Natural Language Processing (NLP)#

NLP enables machines to understand, interpret, and generate human language. Azure AI Language provides comprehensive NLP capabilities.

Core NLP Capabilities#

Conversational AI Services#

Azure AI Speech Services#

5. Generative AI & Azure OpenAI Service (20-25%)#

Generative AI is the highest weighted domain on the exam. It covers creating new content—text, images, code, and more—using Azure OpenAI Service and models like GPT-4, DALL-E, and Whisper.

Generative AI Concepts#

Azure OpenAI Service Models#

Prompt Engineering#

Effective prompting dramatically improves results:

Retrieval Augmented Generation (RAG)#

RAG combines retrieval and generation to ground responses in your data:

1. Index your documents in Azure AI Search
2. Retrieve relevant chunks based on user query
3. Include retrieved content in the prompt
4. Generate grounded response using your data

---

Know your range before test day#

Five AI-900 domains checked. Next: a short diagnostic, free practice on weak topics, then a timed sitting when you want a readiness signal.

• Timed diagnostic with domain-level breakdown
• Clear signal on what to drill next
• Practice free; timed sittings and readiness range on Pro

Take the free AI-900 assessment →

---

Next step: Start practicing free on the AI-900 hub, or sit a short diagnostic first.