The Middle of AI

What it is

My goal is to demystify modern AI models for regular people. This content helps people understand how models work and run them on their own hardware, with an emphasis on the Medium Language Model (MLM) space. "The Middle of AI" is what I'm calling the sum of that content - articles, webinars, this website; it's all here to help people to run their own AI, whether at work or home.

There's a huge gap in how AI is talked about. On one side, the hype tells you models are magic. On the other, the technical material assumes you already know what a transformer is and how quantization works. Most people live in the middle, and that's exactly where the confusion lives.

Our Webinar Series

In partnership with MMD Services, we are putting on a webinar series called "Learning the New World of AI: MLM Bootcamp". I've committed to "at least 5 sessions" to go through MLMs in extreme detail.

MMD is a premier boutique technology staffing firm in Chicago founded by Maria Dubov. Maria and team are fantastic, and when the opportunity came up to work with them on building content for their AI Community Hub, I was honored to be a part of it.

Our goal for the series was simple, but two-fold:

• If you are a leader who is using AI, we aim to give you enough information to make informed decisions around costs, architecture, usage, and rationale for implementing MLMs at your company.
• If you are a tinkerer or someone who wants to build your own AI at home or work, we aim to teach you how to do it. Not in a "stereo instructions" way, but more of a "teach them to fish" way of showing you how you would learn to build it yourself.

Our first session was on July 21st, 2026 and was a success. The first session gave an overview of the space, why it's important, and some general explanations of how models and harnesses work with key terms to know.

The second session was on August 18th, and began the actual deep dive. We went in-depth on models - where to find them, how to read a model card, compression/quants, and file types.

Our next sessions are scheduled for September 8th, 15th, and October 6th. I'm excited for the rest of the topics we have planned - cache, inference, hardware, benchmarking, tuning, and customizations.

You can sign up for future sessions and find recordings of our previous sessions at MMD's AI Community Hub site:

I originally planned to go in-depth on agent harnesses as well, but there's already at least five sessions worth of content here and the likely outcome is we'll set up a separate series for agents.