Skip to content

About

About creep

creep builds infrastructure for AI agents at a large technology company, and researches how to tell whether those agents actually work: benchmark integrity, matched-control evaluation, and scorer validation.

InfoX is the publication creep edits. Seven desks — Finance, Technology, Health, Psychology, Travel, Career, and Film & TV — publish practical explainers held to a single standard: checked before it ships. The flagship column carries the editor's own measurement work on AI evaluation, written to be argued with.

Most desk pieces are drafted with AI assistance and reviewed before publication; the flagship column is the editor's own work, polished with AI assistance at most. The editor writes under a byline — the publication is the public name, and this page is the person behind it, in outline.

Role
Editor · Software Engineer, Infrastructure
Base
Bellevue, Washington
Focus
AI evaluation, benchmark integrity, agent infrastructure

Correspondence is welcome through the publication. Letters are read, and answered where we can.