Local LLMs - Agentic Software Engineering Meetup

​The Agentic Software Engineering meetup is for software engineers and IT professionals who use AI tools in their craft.

​agentic engineering describes the practice of developing software with the assistance of coding agents.

What are coding agents? They're agents that can write and execute code. Popular examples include Claude Code, OpenAI Codex, Gemini CLI, and https://kilo.ai/

If you build software and want to learn and share effective practices for using agents to build real software, this is the meetup for you.

This meetup will focus on Local LLMs and using them for coding, and include discussion and demos of how to run them, the hardware needed, and the pros and cons of doing so.

Agenda:
5:30-6:00pm: networking & setup
6:00pm - Introductions
6:15pm - Evaluating Local LLMs: Accuracy, Benchmarks, and Model Selection
6:45 - Local LLM Demos & State of the art versus data center inference
7:30pm - End

​

​Talk: Evaluating Local LLMs: Accuracy, Benchmarks, and Model Selection

How do you determine whether a local LLM is accurate enough for your task? Drawing on my hands-on experience with local LLMs, I’ll discuss designing task-specific benchmarks, measuring correctness, and comparing models within practical hardware constraints. The session will explore how data structure, task complexity, and retrieval influence performance, with takeaways for developers choosing local models for AI-assisted software development.

Speaker: Maryam Shahbaz Ali is an AI Engineer at Freedom Buckets and a Research Trainee at Children’s National Hospital. Her work focuses on local LLMs, retrieval-augmented generation, and evaluating AI systems for practical applications. She holds an MS in Business Analytics from The George Washington University. Connect at: /maryamshahbazali