Microsoft · 2026 Edition
A complete preparation guide written by Microsoft-certified engineers. Covers the exam format,all 10 blueprint domains, a week-by-week study plan, and proven tips for passing first time.
3–4 months
Prep time
Intermediate
Difficulty
50
Exam questions
700/1000
Pass mark
Exam code
AI-102
Full name
Azure AI Engineer Associate
Vendor
Microsoft
Duration
100 minutes
Questions
50 items
Passing score
700/1000 (scaled)
Domains covered
10 blueprint domains
Recommended experience
1–2 years of software development or data science experience; familiarity with Python or C#
Typical prep time
3–4 months
AI-102 earns the Azure AI Engineer Associate certification. It validates the skills to build, deploy, and integrate Azure AI services into applications — a role in very high demand as organisations implement AI features across their products.
Job roles this opens
Domain percentage weights are not currently available for this exam. The checklist below is still useful for planning your study.
Weeks 1–2
Plan and Manage Azure AI Solutions: multi-service accounts, monitoring, responsible AI
Tip: Azure AI services have two account types: multi-service resource (one key for all services, simplifies billing) and single-service resource (separate key per service, finer-grained access control). Know when each is appropriate.
Weeks 3–5
Computer Vision: Azure AI Vision, Custom Vision, Video Indexer, Face API
Tip: Azure AI Vision capabilities by what they detect: OCR (text in images), image tagging (objects/scenes), object detection (objects with bounding boxes), background removal, and dense captioning. Know the difference between the Read API (multi-page document OCR) and standard OCR for individual images.
Weeks 6–8
Natural Language Processing: Language service, Text Analytics, Translator, CLU
Tip: Azure AI Language service consolidates multiple Cognitive Services APIs. Know each capability: sentiment analysis, entity recognition (NER), key phrase extraction, language detection, question answering, and conversational language understanding (CLU). Questions describe an NLP task and ask which feature to use.
Weeks 9–11
Document Intelligence, Azure AI Search, and Generative AI (Azure OpenAI)
Tip: The RAG pattern on AI-102: user question → vector search against a knowledge base (Azure AI Search) → context injected into the prompt → LLM generates a grounded response. Know the components and why each step is necessary.
AI-102 has lab questions — you may need to complete tasks in the Azure AI Foundry portal, Vision Studio, or Language Studio. Practice in these interfaces before your exam date.
Azure AI Search (formerly Cognitive Search) core concepts: index (structured data store), indexer (pulls data from a source), skillset (AI enrichment pipeline), and search service. AI Search supports vector search for semantic similarity alongside traditional keyword search.
Azure OpenAI deployments: you deploy a model (GPT-4, GPT-4o, text-embedding-ada-002) to an endpoint. RAG is preferred over fine-tuning for grounding responses in current data — fine-tuning teaches style and format, RAG provides current knowledge.
Responsible AI in Azure OpenAI: content safety filters block harmful content by category (hate, violence, sexual, self-harm). Azure AI Content Safety is a separate service for applying these filters outside of OpenAI deployments.
Custom model training: Custom Vision (image classification/detection), Custom Document Intelligence (extraction models for your form layout), and Custom NER (entity extraction on your domain text). Know when to use pre-built vs custom models — custom is warranted when pre-built accuracy is insufficient for your domain.
Apply everything in this guide with adaptive practice questions, detailed answer explanations, and domain analytics.
Deep-dive explanations of the key topics tested on AI-102 — with exam key points and common misconceptions.