Hi HN. I built ZOT as an experiment to see if biological immune systems (Darwinian selection + Kinetic Proofreading) could replace massive ML models for anomaly detection.

It’s a 400+line, zero-dependency Rust binary that reads raw hardware telemetry (memory and clock latency) to autonomously evolve threat receptors. It hits ~95% accuracy in synthetic lab tests with near-zero false positives.

I’ve taken the theory and the initial prototype as far as I can, but I don't have the cycles to maintain this full-time or test it in enterprise environments. I'm releasing the code and the thesis under the MIT license. If anyone wants to fork it, break the math, or test it in the wild, the repo is yours.