AI Developer - Azure AI Foundry and Semantic Kernel Fundamentals
October 1, 2025 ยท View on GitHub
Introduction
The Azure AI Foundry and Semantic Kernel Fundamentals is an introduction to understanding the conceptual foundations of infusing your application with AI using the Semantic Kernel development kit. Materials from this workshop can serve as a foundation for building your own AI infused solutions with Semantic Kernel.
This workshop consists of eight challenges and is designed to encourage learning and research. If you want a deeper understanding of how to implement an AI solution but have little or no experience with OpenAI or Semantic Kernel then this workshop is for you.
We recommend to host this as a team based activity where students work in groups of 3-5 people to solve the challenges. Each challenge is designed to be completed in 30-90 minutes.
Learning Objectives
This workshop is designed for individuals who want to gain practical experience in working with Azure AI Foundry and Semantic Kernel. By applying this knowledge, you will be able to integrate AI into your own applications.
Participants will learn how to:
- Build a simple chat using Semantic Kernel and C#
- Add plugins and enable auto calling to create Planners
- Import existing APIs using OpenAPI
- Implement Retrieval Augmented Generation (RAG)
- Document Chunking
- Grounding AI
- Working with Image generation
- Multi-Agent workflows
Challenges
eShop App
- Challenge 09: Infuse existing Apps with AI
- Enhance the Product Search functionality by implementing AI-powered search capabilities using Semantic Kernel
- Transform traditional keyword-based search into an intelligent Semantic Search system leveraging Azure AI Search Service for more contextual and relevant results
- Challenge 10: Enrich the users search experience using AI
- Implement an AI-powered product description generator that creates compelling sales pitches based on user search queries
- Integrate Semantic Kernel to dynamically generate personalized, context-aware product descriptions that replace standard catalog text
- Challenge 11: Building a "Real World" Multi-Agent Solution
- Implement an AI Agent registry for tracking a collection of agents used to solve our workflow
- Create a Group Chat with Round Robin Orchestration to show a simple baseline
- Challenge 12: Adding Realtime Agent Response Callback
- Implement a callback method that will accept the response from the Agent Chat as it occurs to update the UI
- Enforcing a parsable format to ensure responses are easily used by applications.
- Challenge 13: Build your own MCP Tool
- Build an MCP server in .NET
- Using an MCP server to enhance Agentic workflows
- Challenge 14: Creating a Task Tracking Agent
- Create an Agent that specializes in monitoring and managing a workflow state
- Adding MCP tools to the Agent to give it "permanance"
- Challenge 15: Custom Agent Workflow
- Creating a sophiticated InteractiveGroupChatManager
- Handling decision making for AI coordination
- Challenge 16: Human-in-the-Loop
- Creating Human-in-the-Loop AI Workflows
- Real-time communication between AI Agent workflows and User Applications