Microsoft Azure AI Fundamentals (AI-900) Certification Prep

Microsoft Azure AI Fundamentals

Official preparation course for the Microsoft Azure AI Fundamentals (AI-900) certification exam. O'Reilly live training -- 5 hours, 80% demos, 20% theory.

Exam Information

Exam Code: AI-900
Objectives Updated: May 2, 2025
Official Page: Microsoft Learn AI-900
Practice Assessment: Free official practice questions
Exam Sandbox: aka.ms/examdemo

Course Flow (5 Segments)

| Segment | Topic | Time | Exam Weight | |---------|-------|------|-------------| | S1 | AI Fundamentals & Responsible AI | 09:00 AM | 15-20% | | S2 | Machine Learning | 10:00 AM | 15-20% | | S3 | Computer Vision | 11:00 AM | 15-20% | | S4 | Natural Language Processing | 12:00 PM | 15-20% | | S5 | Generative AI | 01:00 PM | 20-25% (highest) |

For full objective details, see AI-900-exam-objectives.md.

Repository Structure

ai900/
├── .github/                        # GitHub config (CODEOWNERS, templates, Copilot agents)
│   ├── agents/                     # GitHub Copilot agent definitions
│   ├── instructions/               # Copilot custom instructions
│   ├── prompts/                    # Copilot prompt files
│   └── ISSUE_TEMPLATE/
├── bicep/                          # CAF-aligned IaC for lab environment
│   ├── main.bicep
│   ├── modules/                    # ai-services, openai, doc-intel, ML, etc.
│   └── parameters/                 # dev.bicepparam, prod.bicepparam
├── demos/                          # Live class demos (Python + uv)
│   ├── .env.example                # Template for Azure credentials
│   ├── assets/                     # Shared media & datasets (LFS-tracked)
│   │   ├── Audio-Video/
│   │   ├── CSV/
│   │   ├── OCR/
│   │   ├── People/
│   │   ├── Places/
│   │   └── Things/
│   ├── hour-1-ai-fundamentals/     # Vision, Content Safety, Responsible AI
│   ├── hour-2-machine-learning/    # scikit-learn classification/regression/clustering
│   ├── hour-3-computer-vision/     # Image Analysis, OCR, Face, Document Intelligence
│   ├── hour-4-nlp/                 # Sentiment, NER, Speech, CLU
│   └── hour-5-generative-ai/      # GPT-4o, DALL-E 3, prompt engineering, RAG
├── docs/                           # Course materials & exam prep
│   ├── warner-ai900-feb-2026.pptx  # Slide deck (current delivery)
│   ├── AI-900-exam-objectives.md   # Full objective domain
│   ├── AI-900-CORE-RESOURCES.md    # Curated study materials
│   ├── AI-900-PRACTICE-QUESTIONS.md
│   ├── AI-900-PRACTICE-QUESTIONS-SET2.md
│   ├── PRACTICE-QUESTIONS-GUIDE.md # Practice exam resources
│   ├── MCP-DOCS-SERVER-GUIDE.md    # Claude AI + MS Docs for cert prep
│   ├── INDEX.md                    # Guided tour of all materials
│   ├── LEARNING_RESOURCES.md
│   └── exam-metadata/
├── feb-2026/                       # Current delivery course plan
│   ├── course-plan-feb-2026.md
│   └── to-be-archived/            # Legacy content staged for removal
├── images/                         # README cover images
├── practice-questions/             # Practice questions by domain
│   ├── 01-ai-workloads-and-considerations/
│   ├── 02-machine-learning-on-azure/
│   ├── 03-computer-vision-workloads/
│   ├── 04-nlp-workloads/
│   └── 05-generative-ai-workloads/
├── scripts/                        # Utility scripts
│   ├── deploy-ai-services.sh
│   ├── cleanup-ai-services.sh
│   ├── validate-links.py
│   └── github-cli.ps1
└── temp/                           # Temporary working files

Running the Demos

Each demo is a standalone Python project managed with uv. You need Python 3.13+ and uv installed.

cd demos/hour-1-ai-fundamentals
uv sync                   # creates .venv and installs dependencies
uv run python main.py     # launches interactive menu

Repeat for any hour-N-* folder. All demos share a single demos/.env file for Azure credentials:

cp demos/.env.example demos/.env
# Fill in your Azure AI Services keys and endpoints

Bicep Deployment

Deploy the full lab environment with one command:

az deployment group create \
  --resource-group AI900-Feb2026 \
  --template-file bicep/main.bicep \
  --parameters bicep/parameters/dev.bicepparam

Azure Terminology (Current as of May 2025)

The exam uses current Azure service names exclusively. Deprecated names appear only as wrong answers.

| Deprecated Name | Current Name | |-----------------|--------------| | Cognitive Services | Azure AI Services | | LUIS | CLU (Conversational Language Understanding) | | QnA Maker | Custom Question Answering | | AI Studio | Microsoft Foundry (ai.azure.com) | | Language Studio | Deprecated -- use Microsoft Foundry portal |

Key Azure Portals

| Portal | URL | Notes | |--------|-----|-------| | Azure Portal | portal.azure.com | Resource management | | Microsoft Foundry | ai.azure.com | Primary for Language, GenAI, model catalog | | Azure ML Studio | ml.azure.com | AutoML, Designer, endpoints | | Vision Studio | portal.vision.cognitive.azure.com | Image Analysis, OCR, Face | | Speech Studio | speech.microsoft.com | Speech-to-text, text-to-speech | | Custom Vision | customvision.ai | Image classification, object detection |

Study Resources

In This Repo

Practice Questions Guide -- 15 practice exam resources and study strategies
AI-900 Core Resources -- Curated study materials, practice exams, MS Learn paths
MCP Microsoft Docs Server Guide -- Use Claude AI + MS Docs for interactive cert prep
Docs Index -- Guided tour of all course materials

Microsoft Learn Path

Introduction to AI in Azure -- 14-module learning path covering all five exam domains

Register for Exam

Prerequisites

Basic understanding of cloud computing concepts
Microsoft Azure subscription (free trial or paid)
Python 3.13+ and uv for running demos
Interest in artificial intelligence and machine learning

Instructor

Name: Tim Warner
Title: Microsoft MVP & Certified Trainer
Website: techtrainertim.com
LinkedIn: Timothy Warner

License

This course material is licensed under the MIT License. See the LICENSE file for details.

Ai900

Install / Use

README