ExamFinch.com My Study

Cheat sheet

Azure AI-102 Cheat Sheet

Practice this exam

Plan + Manage Solution

  • 20-25% of exam
  • Resources + Endpoints Auth + Security Deploy + Ops Responsible AI Service selection

Generative AI

  • 15-20% of exam
  • Azure OpenAI + Foundry Prompt + Tuning RAG grounding Fine-tuning Image generation (gpt-image)

Agentic Solution

  • 5-10% of exam
  • Agents Tools Foundry Agent Service Agent Framework Multi-agent orchestration

Computer Vision

  • 10-15% of exam
  • Vision Services Image Analysis + OCR Custom Vision Video Indexer Spatial Analysis (retired)

Natural Language Processing

  • 15-20% of exam
  • Language (Text) Speech + CLU Sentiment + PII Translator Question answering

Knowledge Mining

  • 15-20% of exam
  • AI Search Doc Intelligence Vector + hybrid search Content Understanding Skillsets

Quick Facts

  • Exam AI-102
  • Credential Azure AI Engineer
  • Time 100 min
  • Pass 700/1000
  • Questions Not published
  • Fee $165 USD
  • Blueprint Dec 23 2025
  • Retired June 30 2026

Six Skill Areas PL

  • Plan | GenAI | Agent | Vision | NLP | Mining
  • Plan: 20-25% GenAI: 15-20% Agent: 5-10% Vision: 10-15% NLP: 15-20% Mining: 15-20%

Multi vs Single Service PL

Multi-service

  • One endpoint/key
  • Shared billing
  • Fast start

Single-service

  • Dedicated endpoint
  • Per-service RBAC
  • Network isolation
  • Simple vs isolated

Service Picker PL

  • Generate text/images → Azure OpenAI ( Generative )
  • Analyze images → Azure AI Vision ( Prebuilt )
  • Own image classes → Custom Vision ( Trained )
  • Extract form fields → Document Intelligence ( Structured )
  • Search your content → Azure AI Search ( Retrieval )
  • Transcribe audio → Azure AI Speech ( STT )
  • Detect user intent → CLU ( Language )
  • Multimodal extraction → Content Understanding ( Foundry Tools )

Resources + Endpoints PL

  • Multi-service resource One key many services
  • Single-service resource Dedicated endpoint per service
  • Azure OpenAI resource Own kind OpenAI
  • Endpoint Service base URL
  • Key pair Two rotatable keys
  • Region Fixed by resource
  • kind Resource type flag

Responsible AI Layers PL

  • Filters | Blocklists | Shields | Harm detection
  • Filters: severity levels Blocklists: exact terms Shields: jailbreak block Harm: category scores

Key vs Managed Identity PL

API key

  • Shared secret
  • Manual rotation
  • Leak risk

Managed identity

  • No secrets
  • Entra tokens
  • RBAC scoped
  • Secret vs identity

Auth Picker PL

  • Azure-hosted app → Managed identity ( No secrets )
  • Local dev + prod → DefaultAzureCredential ( Chained )
  • Grant resource access → Azure RBAC ( Roles )
  • Private clients only → Private endpoint ( Disable public )
  • Getting 429 → Backoff retry ( Check quota )

Auth + Security PL

  • Managed identity No stored secrets
  • DefaultAzureCredential Chained auth methods
  • Microsoft Entra Token-based identity
  • Azure RBAC Role-based authorization
  • Private endpoint Private network only
  • Disable public access Close public endpoint
  • 429 throttle Retry exponential backoff

Content Safety vs Filters PL

Content Safety

  • Standalone service
  • Any content
  • Text and images

OpenAI filters

  • Built into deployment
  • Prompt/completion only
  • Not external content
  • Anywhere vs model-only

Deploy + Ops PL

  • IaC Repeatable resource templates
  • CI/CD Automated deployment pipeline
  • Container Local/edge deployment
  • Azure Monitor Metrics and logs
  • Diagnostic settings Route resource logs
  • Cost management Track and budget
  • Quota Rate and token limits

Responsible AI PL

  • Content Safety Standalone content moderation
  • Content filters Model prompt/completion filter
  • Blocklists Exact banned terms
  • Harm categories Hate sexual violence self-harm
  • Severity thresholds Per-category block levels
  • Prompt Shields Block jailbreak attacks
  • Model card Intended use limitations

Generative Flow GE

  • Deploy | Prompt | RAG | Evaluate | Optimize
  • Deploy: model Prompt: template RAG: ground Evaluate: score Optimize: tune

Fine-tune vs RAG GE

Fine-tune

  • Change behavior/style
  • Labeled JSONL
  • Slower iteration

RAG

  • Add fresh grounding
  • No retraining
  • Cheaper iteration
  • Behavior vs knowledge

Azure OpenAI + Foundry GE

  • Foundry hub Shared compute and connections
  • Foundry project One solution workspace
  • Azure OpenAI GPT gpt-image embeddings
  • Standard deployment Pay per token
  • Provisioned throughput Reserved predictable capacity
  • gpt-image Image generation (DALL-E 3 retired)
  • Embedding model Text to vectors

Prompt + Tuning GE

  • Prompt flow Pipeline tool; retires April 2027
  • Prompt template Reusable parameterized prompt
  • System message Durable role instructions
  • Temperature Controls randomness
  • Fine-tuning Train on examples
  • Evaluation Score models and flows
  • Grounding Answer from retrieved sources

Agents AG

  • Agent Reasoning plus tools
  • Tool Callable agent capability
  • Foundry Agent Service Managed agent hosting
  • Agent Framework Code-first complex agents
  • Supervisor pattern Routes specialist agents
  • Orchestration Coordinate multi-agent workflow
  • Guardrails Max turns limits

Classification vs Detection VI

Classification

  • What is present
  • Whole-image label
  • No boxes

Object detection

  • What and where
  • Bounding boxes
  • Location matters
  • What vs where

Vision Services VI

  • Image Analysis Prebuilt tags and captions
  • OCR / Read Extract printed handwritten text
  • Custom Vision Train your images
  • Classification What is present
  • Object detection What and where
  • Compact domain Exportable edge model
  • Video Indexer Video insights and transcripts
  • Spatial Analysis Retired March 30, 2025

CLU vs Question Answering NL

CLU

  • Intents + entities
  • Bounded actions
  • Trained utterances

Question answering

  • Knowledge base
  • FAQ + docs
  • Best answer match
  • Action vs answer

Language (Text) NL

  • Sentiment analysis Positive negative neutral mixed
  • Key phrase extraction Main text points
  • NER Detect named entities
  • PII detection Find redact sensitive data
  • Language detection Identify text language
  • Translator Text and document translation

Speech + CLU NL

  • Speech-to-text Transcribe spoken audio
  • Text-to-speech Synthesize spoken output
  • SSML Control pronunciation and prosody
  • Custom Speech Adapt domain vocabulary
  • CLU Intents and entities
  • Question answering Knowledge base answers
  • Custom Translation Domain-tuned translator model

Search Stack KN

  • Index | Indexer | Skillset | Query
  • Index: store Indexer: ingest Skillset: enrich Query: retrieve

Keyword vs Vector Search KN

Keyword

  • Exact terms
  • Lexical BM25
  • Literal match

Vector

  • Embedding similarity
  • Semantic meaning
  • Handles synonyms
  • Literal vs meaning

RAG + Search Picker KN

  • Ground model answers → RAG pattern ( Retrieve + prompt )
  • Exact keyword match → Lexical search ( BM25 )
  • Meaning similarity → Vector search ( Embeddings )
  • Best default search → Hybrid + semantic ( Combined )
  • Persist enrichments → Knowledge store ( Projections )
  • Enrich while indexing → Skillset ( OCR + phrases )

AI Search KN

  • Index Searchable document store
  • Indexer Pull and populate
  • Skillset Enrich during indexing
  • Knowledge store Persist enriched projections
  • Vector field Embedding similarity search
  • Hybrid search Keyword plus vector
  • Semantic ranker Rerank by meaning
  • Vectorizer Query text to vector

Doc Intelligence KN

  • Prebuilt model Invoice receipt layout ID
  • Custom model Train your forms
  • Composed model Route multiple submodels
  • Layout model Tables structure and text
  • Content Understanding Multimodal schema extraction
  • Knowledge mining Search across documents

Common Traps

Resource type

  • Foundry includes OpenAI ≠ Standalone OpenAI optional

Auth method

  • Keys are secrets ≠ Managed identity tokenless

Moderation scope

  • Content Safety anywhere ≠ Filters model-only

Vision task

  • Classification labels image ≠ Detection locates objects

Language task

  • CLU classifies intent ≠ QnA matches answer

Search type

  • Keyword matches terms ≠ Vector matches meaning

Grounding vs training

  • RAG adds context ≠ Fine-tune changes behavior

Last Minute

  • 1. Pass = 700 of 1000
  • 2. 100 minutes; question count not published
  • 3. Six weighted skill areas
  • 4. Foundry resource includes OpenAI
  • 5. Managed identity over keys
  • 6. Content Safety moderates anywhere
  • 7. RAG grounds, fine-tune shapes
  • 8. Classification labels; detection locates
  • 9. CLU intents; QnA answers
  • 10. Hybrid plus semantic ranker
  • 11. Prompt Shields block jailbreaks
  • 12. Prebuilt before custom models

Azure AI-102

  • Practice Questions 200 questions Study Guide 37 topics Cheat Sheet 84 facts Flashcards 50 cards Videos 2 videos Blogs 2 blogs Podcast