How bartczernicki/decisionintelligence.genai.workshop works
Project type and purpose
• The project is an interactive workshop course designed to demonstrate and experiment with Decision Intelligence concepts using Generative AI.
• Its purpose is to guide diverse users (from non-technical readers to AI engineers) through decision-making frameworks by integrating interactive Jupyter notebooks (via VS Code and .NET Interactive), C# plugins, ML models, and generative AI orchestration (Semantic Kernel, OpenAI).
File structure
• Top-level directories include:
• Classes: Contains C# code such as “BaseballHallOfFamePlugin.cs,” which likely extends decision-making functionality.
• Data: Hosts datasets (e.g., the CSV file for MLB batters) used in decision examples.
• Models: Contains ML model files (e.g., an ML.NET file for a generalized additive model) that might be used to support decision analysis.
• Notebooks: Contains all the interactive workshop modules written as Jupyter notebooks that guide users through decision intelligence workflows. They range from introductory decision-making frameworks to deep dives using Semantic Kernel and OpenAI.
• README: Describes the workshop methodology, prerequisites, and module details.
• The structure reflects a modular, layered design where educational content (Notebooks) ties together practical code (Classes), data/model resources (Data, Models), and supporting visuals (Images).
README for additional insights
• The README reveals several personas (read, technical, Decision Intelligence AI Engineer), indicating that the design should accommodate both interactive learning (notebooks) and extension points (modifying C# code, decision prompts, and integration with AI APIs).
• There is a significant emphasis on interactive execution in VS Code using .NET Interactive and requirements like Azure OpenAI and Semantic Kernel orchestration.
• The workshop modules are organized sequentially, offering both a conceptual and hands-on approach to decision intelligence.
Final advice
• Be as detailed as possible, ensuring every relevant component is represented.