Study Notes

Appearance

AI-102 Exam Objectives (2025 Update) ​

The AI-102 exam was restructured in late 2024/early 2025 to emphasize Generative AI and Agentic solutions. The current domains and weights are:

1. Plan and manage an Azure AI solution (20-25%) ​

  • Select the appropriate Microsoft AI Foundry Services.
  • Plan and manage Microsoft AI Foundry hubs and projects.
  • Manage costs for Microsoft AI Foundry Services.
  • Manage security for Microsoft AI Foundry hubs and projects.
  • Integrate Microsoft AI Foundry Services into a CI/CD pipeline.
  • Plan and manage diagnostic settings and alerts.
  • Design a responsible AI governance framework.
  • Implement AI solutions responsibly (Content Safety, Prompt Shields).

2. Implement generative AI solutions (15-20%) ​

  • Select foundational models from the Model Catalog.
  • Implement and manage Large Multimodal Models (LMMs).
  • Implement and customize prompt flows.
  • Evaluate generative AI model performance.
  • Configure prompt engineering and fine-tuning.
  • Integrate model reflection and feedback collection.

3. Implement an agentic solution (5-10%) ​

  • Implement an agent using Microsoft AI Agent Service.
  • Implement tools and code interpreters.
  • Orchestrate multi-agent interactions.
  • Test and deploy agentic solutions.

4. Implement computer vision solutions (10-15%) ​

  • Analyze images using Azure Vision in Foundry Tools.
  • Extract insights from video and live streams.
  • Analyze movement using spatial analysis.
  • Detect and recognize faces.
  • Implement Custom Vision classification and object detection.

5. Implement natural language processing solutions (15-20%) ​

  • Analyze text using Language services.
  • Implement Conversational Language Understanding (CLU).
  • Implement Custom Question Answering.
  • Process speech and implement translation.
  • Integrate generative AI speaking capabilities.

6. Implement knowledge mining and information extraction (15-20%) ​

  • Implement an enrichment pipeline in Azure AI Search.
  • Implement vector and hybrid search.
  • Implement integrated vectorization.
  • Implement Document Intelligence models.
  • Implement Azure Content Understanding OCR pipelines.