Description
Mindrift is seeking experienced software engineers to design evaluation tasks for frontier AI coding agents, focusing on building realistic developer environments, creating tempting but unsafe shortcuts, and writing tests that verify agents complete tasks safely and within scope. The role is project-based rather than permanent employment, requires strong software development and test design experience with Python and JavaScript/TypeScript, familiarity with coding agents and GitHub PR/CI workflows, and English proficiency at B2+; contributors can earn up to $50 per hour equivalent during active phases.
