AI-102: Designing and Implementing a Microsoft Azure AI Solution [120 Minute Read]

---

Welcome & Introduction

Welcome to the MindMesh Academy comprehensive study guide for the Microsoft AI-102: Designing and Implementing a Microsoft Azure AI Solution certification exam.

This guide follows a first-principles pedagogical approach—we build foundational mental models before diving into specific services and implementations. This approach enables you to reason through unfamiliar exam scenarios rather than relying solely on memorization.

Implementation Focus: AI-102 is an implementation exam. Each section includes a 🔧 Implementation Reference box with the SDK classes, methods, headers, and parameters you need to know.

Certification Level: Associate
Skills Measured As Of: April 30, 2025
Passing Score: 700 out of 1000
Exam Duration: 120 minutes
Question Count: 50-55 questions
Question Types: Multiple choice, multiple select, case studies, drag-and-drop, code completion

---

Exam Purpose & Audience

The AI-102 exam validates your ability to build, manage, and deploy AI solutions using Azure AI services. From the official description:

"Candidates should be able to use REST APIs and SDKs to build secure image processing, video processing, natural language processing, knowledge mining, and generative AI solutions on Azure."

The AI-102 exam covers six domains with the following approximate weights:

---

(Table of Contents - For Reference)

• Phase 1: First Principles of Azure AI Solutions
  • 1.1. The AI Capability Spectrum
  • 1.2. The Microsoft AI Platform Architecture
  • 1.3. The Input-Output Framework
  • 1.4. The Build vs. Buy Decision Framework
  • 1.5. Responsible AI as a Foundation
  • 1.6. Reflection Checkpoint
• Phase 2: Plan and Manage Azure AI Solutions (15-20%)
  • 2.1. Select the Appropriate Azure AI Services
    • 2.1.1. Pre-built AI Services (Foundry Tools)
    • 2.1.2. Azure OpenAI Service Selection (Foundry Models)
  • 2.2. Plan, Create and Deploy Azure AI Services
    • 2.2.1. Resource Types
    • 2.2.2. Network Configuration
    • 2.2.3. Deployment and Provisioning
  • 2.3. Manage, Monitor, and Secure Azure AI Services
    • 2.3.1. Authentication Methods
    • 2.3.2. RBAC and Access Control
    • 2.3.3. Monitoring and Diagnostics
    • 2.3.4. Cost Management
    • 2.3.5. Throttling and Rate Limits
  • 2.4. Implement Content Safety
  • 2.5. Reflection Checkpoint
• Phase 3: Implement Generative AI Solutions (15-20%)
  • 3.1. Build Solutions with Azure AI Foundry
  • 3.2. Use Azure OpenAI to Generate Content
    • 3.2.1. Chat Completions
      • 3.2.1.1. Message Structure and Roles
      • 3.2.1.2. Sampling and Generation Parameters
      • 3.2.1.3. Response Control and Formatting
      • 3.2.1.4. Context and Token Management
    • 3.2.2. Image Generation (DALL-E)
    • 3.2.3. Embeddings
  • 3.3. Implement RAG Solutions
    • 3.3.1. Vector Embeddings for RAG
    • 3.3.2. Retrieval and Grounding
  • 3.4. Prompt Engineering and Fine-Tuning
  • 3.5. Prompt Templates
  • 3.6. Enable Tracing and Collect Feedback
  • 3.7. Deploy Containers for Local and Edge Devices
  • 3.8. Reflection Checkpoint
• Phase 4: Develop AI Agents (10-15%)
  • 4.1. Understanding AI Agents
    • 4.1.1. Agent Architecture Concepts
    • 4.1.2. Agent Frameworks
  • 4.2. Azure OpenAI Assistants
    • 4.2.1. Assistants and Threads
    • 4.2.2. Built-in Tools
    • 4.2.3. Run Management
  • 4.3. Semantic Kernel Agents
  • 4.4. Autogen Multi-Agent Systems
  • 4.5. Multi-Agent Orchestration Patterns
  • 4.6. Reflection Checkpoint
• Phase 5: Implement Computer Vision Solutions (15-20%)
  • 5.1. Analyze Images
    • 5.1.1. Visual Feature Analysis
    • 5.1.2. OCR and Text Extraction
    • 5.1.3. Object and People Detection
  • 5.2. Implement Custom Vision Models
    • 5.2.1. Image Classification
    • 5.2.2. Object Detection Projects
  • 5.3. Analyze Videos
  • 5.4. Reflection Checkpoint
• Phase 6: Implement Natural Language Processing Solutions (15-20%)
  • 6.1. Analyze and Translate Text
    • 6.1.1. Text Analytics Features
    • 6.1.2. Translation Services
  • 6.2. Process and Translate Speech
    • 6.2.1. Speech-to-Text (STT)
    • 6.2.2. Text-to-Speech (TTS)
    • 6.2.3. SSML (Speech Synthesis Markup Language)
    • 6.2.4. Custom Speech and Word Error Rate (WER)
  • 6.3. Implement Custom Language Models
    • 6.3.1. Conversational Language Understanding (CLU)
      • 6.3.1.1. Intents and Utterances
      • 6.3.1.2. Entity Extraction
      • 6.3.1.3. CLU Training and Deployment
    • 6.3.2. Custom Question Answering
    • 6.3.3. Multi-Language Question Answering
  • 6.4. Reflection Checkpoint
• Phase 7: Implement Knowledge Mining and Document Intelligence (15-20%)
  • 7.1. Implement Azure AI Search
    • 7.1.1. Search Fundamentals
      • 7.1.1.1. Index Schema Design
      • 7.1.1.2. Indexers and Data Sources
      • 7.1.1.3. Query Syntax and Execution
    • 7.1.2. Skillsets (AI Enrichment)
    • 7.1.3. Knowledge Store Projections
  • 7.2. Implement Document Intelligence
    • 7.2.1. Pre-built Models (Invoice and Receipt)
    • 7.2.2. Custom Models
    • 7.2.3. Layout Analysis
    • 7.2.4. Identity and Other Pre-built Models
  • 7.3. Extract Information with Azure AI Content Understanding
  • 7.4. Reflection Checkpoint
• Phase 8: Exam Readiness
  • 8.1. Exam Structure and Question Patterns
  • 8.2. Common Exam Traps and Distinctions
  • 8.3. Decision Framework Quick Reference
  • 8.4. Implementation Quick Reference
  • 8.5. Practice Questions (71 Questions)
• Phase 9: Comprehensive Glossary
• Phase 10: Conclusion