Nunug

AI can write code, but can it review a real .NET codebase intelligently? Simply handing an LLM a pull-request diff often produces shallow results because the model doesn't understand the surrounding C# code, application architecture, team standards, or the intent behind the change. To make AI genuinely useful for development work, we need to give it more than a prompt—we need to give it context and tools.

In this session, we'll walk through the architecture of a real AI-powered code review system built for .NET development using local language models and Model Context Protocol (MCP) tools. We'll explore how an AI agent can investigate a repository, search and inspect related code, incorporate static-analysis results, apply organization-specific coding standards, and gather the context it needs before making recommendations. Along the way, we'll discuss practical lessons about local models, context engineering, MCP and tool calling, and what it takes to move from an interesting AI experiment to a developer tool that can actually understand and review a real-world codebase.
This is a hybrid meeting!

Join us in-person (come fed - food will not be provided):

Mindfire Office
872 W Heritage Park Blvd STE 200
Layton, UT

Join us online (MS Teams):
https://link.nunug.org/nunug-online-meeting

Learn more at nunug.org!