Ken Rapko

I started out as a mechanical engineer at Virginia Tech, drifted toward robotics and computer vision, and ended up writing software for a living. Five years at Lockheed Martin doing CV and ML R&D — multithreaded vision pipelines, geospatial toolkits, and a research project on Counter-Factual Regret Minimization that earned $100K of internal funding. A master's at Georgia Tech in computational perception along the way.

Now I'm at Nike, where I led the metadata pipeline for Google Search (DSPy-tuned prompts, +16% conversion in pilot) and run point on the team's MLOps infrastructure — Jenkins, Databricks Asset Bundles, AWS SAM, the kind of plumbing that makes ML actually reach production.

Outside of work I write at kenrapko.substack.com, mostly about sports analytics and prediction markets, and maintain a data platform for the Colorado Laboratory to Combat Human Trafficking.