Microsoft Certified: Azure AI Fundamentals (AI-900)

Business units are already asking IT about AI, and many teams lack a shared vocabulary for what Azure AI services do, how usage affects planning, and how to use them responsibly. This AI-900 course gives IT Directors and Team Leads a common baseline before AI projects reach production.

Role fit: The early skills suit managers, cloud administrators, and operations staff with no prior AI experience. They cover machine learning workloads, common AI workload concepts, and Microsoft's six responsible AI principles. Later skills are hands-on with Python, PowerShell, SDKs, and Docker, and they assume basic Python familiarity.

Time investment: The full path runs about 19 hours across 20 skills. Developers should complete it end to end. Non-coders can focus on the concept skills and the no-code Designer pipelines for regression and classification.

Risk and governance: The course covers responsible AI, the do's and don'ts of handling facial and other sensitive personal data, and model reliability and safety. These topics help teams set guardrails and plan governance before adopting AI tools.

Change management: Use this course as the shared foundation for AI adoption so that decision-makers and builders speak the same language.

Microsoft retired the AI-900 exam on 30 June 2026, so the course is assigned today for the shared Azure AI baseline it builds. Team reporting tracks each learner's progress and completion.

When business units start asking for AI, IT teams that complete this training can answer with working prototypes and clear guardrails instead of guesses. The scenarios come straight from the course: no-code regression and classification pipelines built by drag-and-drop in the Azure Machine Learning Designer, Custom Vision object detection models trained on tagged images, and Computer Vision Read OCR through Azure Cognitive Services, including containerized runs.

• Faster, lower-risk prototyping: Your team can stand up an Azure Machine Learning workspace, run AutoML jobs, and pick real-time or batch endpoints based on how applications will use the predictions. That means proof-of-concept work without building everything from scratch.
• Practical document automation groundwork: Practitioners work with Computer Vision OCR capabilities and containerized Cognitive Services, giving Team Leads a clearer path for testing text extraction workflows before broader deployment.
• Governance before production: Microsoft's six responsible AI principles, along with practical do's and don'ts for facial recognition and other sensitive personal data, help your team decide how much model accuracy is enough and how to handle private data.
• Deployment-ready integration skills: Developers call Cognitive Services with Python, PowerShell, and SDKs, and run services in Docker containers. They also script container builds and pushes to Azure Container Registry, so AI capabilities fit into existing DevOps pipelines that move from dev to test to production.

Knowledge

• How AI and machine learning differ, and how models learn from inputs and outputs to predict new results
• Common AI workloads, including object detection, computer vision, natural language processing, and speech, and where Azure Cognitive Services fits
• How Azure Machine Learning Studio organizes data, jobs, components, pipelines, and environments across dev, test, and production
• The difference between real-time and batch endpoints, and when to use each
• Microsoft's six responsible AI principles, plus do's and don'ts for handling facial and other sensitive personal data
• How deployed Azure ML models expose HTTP endpoints for use by other applications

Ability

• Train a basic predictive model in Python by separating inputs from outputs
• Run AutoML jobs and build no-code regression and classification pipelines in the Azure ML Designer
• Tag images with accurate bounding boxes and varied samples, then train Custom Vision models
• Call Cognitive Services computer vision, facial recognition, and OCR APIs with Python, PowerShell, and SDKs
• Package and deploy Cognitive Services in Docker containers and push images to Azure Container Registry
• Publish and deploy an Azure ML model as a real-time endpoint for application consumption