Browse Certification Practice Tests by Exam Family

Microsoft AI-102 Cheat Sheet: Azure AI Engineer

May 1, 2026

Review the legacy Microsoft Azure AI Engineer (AI-102) route, AI service selection, responsible AI, vision, language, search, and the current AI-103 replacement path.

On this page

AI-102 is an older Azure AI Engineer route. Use this cheat sheet as a transition map: keep transferable Azure AI service concepts, but verify whether AI-103 is now the right target for your exam plan.

Use this as a route check. Review the AI-102 scope, then compare the current AI-103 Azure AI apps-and-agents page before studying deeply.

Open AI-102 exam page Compare AI-103

Exam snapshot

FieldDetail
IssuerMicrosoft
Legacy routeAzure AI Engineer Associate
Exam codeAI-102
Current statusReplacement guidance
Closest current examAI-103 Azure AI Apps and Agents Developer Associate
IT Mastery statusExam-selection sample question page

Transition map

Older AI-102 areaWhat still mattersCurrent-route trap
Azure AI servicesMatch language, vision, document, search, and speech capabilities to workload needsStudying service recognition without app and agent integration
Responsible AIGrounding, content controls, evaluation, monitoring, and human reviewTreating safety as only a content filter
Computer visionImages, detection, classification, OCR, labeling, and validationIgnoring data quality and validation examples
Language and searchText analysis, extraction, search indexes, retrieval, and rankingConfusing search relevance with model reasoning
Generative AIRAG, prompts, grounding, citations, evaluation, and operational monitoringAssuming older AI-service patterns cover agent workflows fully

Must-know distinctions

DistinctionHow to decide
AI-102 vs AI-103AI-102 is the older AI Engineer route; AI-103 is the newer apps-and-agents direction.
Service selection vs solution designService selection names the tool; solution design connects it to data, security, monitoring, and user workflow.
Extraction vs generationExtraction pulls structured facts from content; generation creates new language output.
Grounding vs trainingGrounding supplies context at response time; training changes model behavior through data and optimization.
Evaluation vs monitoringEvaluation tests behavior before or during release; monitoring observes production behavior over time.

High-yield checklist

  • Confirm whether AI-102 is still the exam you can actually schedule.
  • Map older AI-service knowledge to AI-103 apps, agents, and Microsoft Foundry concepts.
  • Know when document intelligence, computer vision, language, search, or generative AI is the best fit.
  • Always connect AI output quality to grounding, evaluation, content controls, and monitoring.
  • Use human review when low-confidence extraction or sensitive decisions are involved.
  • Treat identity, data access, and logging as part of the AI architecture.

Common traps

  • Studying old AI-102 notes without comparing the current AI-103 route.
  • Choosing a generative model for a structured extraction problem.
  • Treating citations as optional when unsupported answers are the risk.
  • Ignoring security and permissions around retrieved data.
  • Overlooking evaluation after a prompt or retrieval change.

Practice strategy

Use the AI-102 exam page to test older scope recognition, then move to AI-103 if your goal is current Azure AI apps and agents. If you miss service-selection questions, drill workload-to-service mapping before practicing broader AI architecture.

Revised on Monday, May 25, 2026