I teach
machines
how to play.
Hi, I'm Michael "Kelby" Kelbert — first-gen college student, ex-neuroscientist, now an AI & Robotics MS student at U-M Flint. I build small, opinionated systems that lean into play. Sometimes they help. Sometimes they're just funny. I think both count.
§02SELECTED things
or use arrow keys
WaterWorks
An ethical game about resource scarcity and the small choices that compound. Built with Godot, branching dialogue, and a guilt meter that's surprisingly hard to game.
Emotion Recognition
Real-time facial emotion classification from webcam OR uploaded image. Live preview through OpenCV, friendly UI through Gradio, model trained on FER-2013.
Movie Recommender
A movie recommender that doesn't just push the obvious. Item-based collaborative filtering on MovieLens, content embeddings as a fallback, served behind a tiny Flask API.
From neurons to networks.
I started in cellular & developmental neuroscience at Michigan State, taking apart how we learn. Now I'm at U-M Flint pursuing an MS in AI with a concentration in Robotics & Human-Centered AI, taking apart how silicon pretends to.
I'm a first-generation college student from Lansing. I taught myself half of what I know on accident. The other half I learned on purpose — usually because I wanted to understand why something worked the way it did.
Outside the IDE: long walks, cheap coffee, and games where the systems are trying to teach you something. If you want to chat about robots, ethics, or anything in between — I'm in.
