Azure AI Engineer Associate (AI-102) prep, adaptive plan with ARIA
AI-102 is Microsoft's intermediate-tier cert for engineers building AI solutions on Azure. About 100 minutes, 40 to 60 questions, 700 of 1000 passes. The 2024 refresh added generative AI, agentic solutions, and Azure AI Foundry, so older material misses about a third of the current weight. AI-102 pairs with AZ-204 for the developer-AI track.
What the exam is
Seven domains, scenario-based multiple choice and multiple response. Code shows up at the API and SDK level (Python or C#), not the model-training level.
| Domain | Weight |
|---|---|
| Plan and manage an Azure AI solution | 15-20% |
| Implement generative AI solutions | 10-15% |
| Implement agentic solutions | 5-10% |
| Implement computer vision solutions | 15-20% |
| Implement natural language processing solutions | 15-20% |
| Implement knowledge mining and information extraction solutions | 15-20% |
| Implement Azure AI Foundry solutions | 5-10% |
The weights look balanced, but the newest three areas (Generative AI, Agentic Solutions, AI Foundry) are where the exam separates current candidates from older syllabi.
How 
ARIA preps you for it
The CAT evaluation produces a per-domain skill estimate before any plan is written. Most candidates land heavy on Generative AI and Agentic Solutions because those areas are newest and community study material is thinnest. A candidate already shipping RAG apps skips prompt basics and spends the time on Foundry orchestration and content safety.
The error backlog tags every wrong answer by Azure AI service (Azure OpenAI, AI Search, Document Intelligence, Speech, Vision, Language, Content Safety, AI Foundry). When misses cluster under one service, I pull it forward until the routing is reflexive.
Common pitfalls
- Studying pre-2024 AI-100/AI-102 material. The 2024 refresh changed the exam significantly. Use Microsoft Learn modules dated 2024 or later, not older books.
- Confusing Azure OpenAI with Azure AI Foundry. OpenAI is a model service. AI Foundry orchestrates models, agents, and content safety. Several questions hinge on the distinction.
- Underestimating Content Safety questions. Microsoft is pushing responsible AI hard, so expect questions on filtering, abuse monitoring, and prompt injection mitigation.
Prerequisite
Microsoft recommends one to two years of software development experience and familiarity with Python or C#, plus AZ-900 or equivalent Azure background. No formal cert prerequisite, but candidates without Azure exposure struggle on the planning and security domains.
Pass guarantee
Same five conditions as every ClaudeLab cert: every milestone done, two mocks at 70 plus, one gauntlet at 80 plus, readiness 80, exam sat inside the 60-day window. Complete the roadmap and fail, and the Exam Ready plan is refunded in full.