Artificial Intelligence
Learn AI
Understand generative AI, RAG, prompt engineering, evaluations, and responsible production patterns.
Start with the first guideYour outcome
Move from AI concepts to a working, evaluated application.
New foundation lesson
Types of machine learning
Compare supervised, unsupervised, and reinforcement learning with practical examples and a downloadable 18-slide presentation.
Core curriculum
The skills to practice
Guided path
AI foundations for cloud professionals
Learn the vocabulary, model lifecycle, prompting patterns, and evaluation habits that make later projects easier.
Suggested pace: 4 weeks · Capstone: AI support assistant
0 of 5 complete
- 01
Understand the model lifecycle
Explain training, inference, and the difference between predictive and generative tasks.
- 02
Select a model for a real task
Write requirements and compare candidates against a small test set.
- 03
Design reliable instructions
Define the output contract and test ambiguous or unsafe inputs.
- 04
Ground answers in evidence
Sketch the retrieval flow and measure retrieval separately from generation.
- 05
Evaluate the complete system
Record quality, safety, latency, and cost results for the support assistant.
Practical guides
Learn through decisions and builds.
Build an Azure RAG project in 7 days
Build a portfolio-ready, citation-backed knowledge assistant with a seven-day plan for retrieval, grounding, evaluation, security, and documentation.
Build a production-minded RAG application on Azure
Design retrieval, citations, evaluation, and security for a useful Azure-based knowledge assistant.
A production prompt engineering checklist
Turn informal prompts into versioned, testable application assets with clear inputs, outputs, and failure behavior.
How to choose an AI model in Microsoft Foundry
Choose a Microsoft Foundry model by task fit, representative evaluations, region and deployment availability, data requirements, latency, and total workload cost.
RAG vs fine-tuning: a practical decision guide
Decide whether prompting, retrieval-augmented generation, fine-tuning, or a combination best addresses changing knowledge, repeatable behavior, evaluation, and operating constraints.