AI Agents Hackathon

Welcome to the AI Agents Hackathon! This event is designed to help you learn about AI Agents, understand how they work, and build your own agentic AI solutions using modern frameworks.

What You'll Learn

• The fundamentals of AI Agents and how they differ from traditional AI solutions
• How to use popular Agentic AI frameworks (AutoGen, Semantic Kernel, Azure AI Agent Service)
• Practical prompt engineering and evaluation techniques
• How to build single and multi-agent systems that can accomplish complex tasks
• Integrating AI Agents with other services like Azure AI Search
• Creating AI solutions that can use tools, reason through problems, and collaborate

Hackathon Structure

This hackathon consists of 7 progressive challenges that will take you from the basics of AI Agents to building sophisticated multi-agent systems:

• Challenge 1: Introduction to AI Agents
• Challenge 2: Exploring Agentic AI Frameworks
• Challenge 3: Fundamentals of Prompt Engineering & Evaluation
• Challenge 4: Functions & Plugins with Semantic Kernel
• Challenge 5: Azure AI Agents SDK - Tools & Agents Simplified
• Challenge 6: Tool Usage & Agentic RAG
• Challenge 7: Multi-Agent Collaboration with Semantic Kernel & AutoGen

Prerequisites

• Basic programming knowledge (Python is preferred)
• Azure subscription (for Azure AI services)
• GitHub account

Getting Started

1. Fork this repository to your GitHub account:

   • Click the "Fork" button at the top right of this repository
   • Select your account as the destination
   • Wait for the forking process to complete

2. Open a GitHub Codespace from the forked repository:

   • Click the "Code" button on your forked repository
   • Select the "Codespaces" tab
   • Click "Create codespace on main"

   Alternatively, click the "Open in GitHub Codespaces" button above to create a codespace directly.

   

3. Progress through each challenge sequentially, starting with Challenge 1

4. Each challenge has its own folder with a README.md containing instructions

Challenge Overview

Challenge 1 - Introduction to AI Agents

Learn what makes AI systems "agentic," understand the theory behind AI Agents, and explore their capabilities and limitations.

Challenge 2 - Exploring Agentic AI Frameworks

Get hands-on with popular frameworks including AutoGen, Semantic Kernel, and Azure AI Agent Service. Understand their architecture and when to use each.

Challenge 3 - Fundamentals of Prompt Engineering & Evaluation

Master the art of writing effective prompts, use Prompty for interactive development, and apply Microsoft's tools to evaluate prompt quality.

Challenge 4 - Functions & Plugins with Semantic Kernel

Learn how to extend AI capabilities by creating custom plugins, exposing functions to AI, and enabling more sophisticated interactions.

Challenge 5 - Azure AI Agents SDK - Tools & Agents Simplified

Explore Azure AI Agents SDK to create conversational agents with varying levels of capability, implementing RAG using file search tools.

Challenge 6 - Tool Usage & Agentic RAG

Implement Retrieval-Augmented Generation with an agentic approach, allowing AIs to search and use information in documents to accomplish tasks.

Challenge 7 - Multi-Agent Collaboration

Build a system where multiple specialized agents work together to solve complex problems, coordinating their efforts toward a common goal.

Resources

• AutoGen Documentation
• Semantic Kernel Documentation
• Azure AI Agent Service
• Prompt Engineering Guide

Contributing

This hackathon is open source! If you find issues or want to improve the materials, please submit a pull request.

License

MIT License