Description
Mindrift is seeking experienced software engineers to join a project-based AI opportunity focused on creating evaluation tasks for frontier coding agents. The work involves building realistic developer environments, designing tasks with unsafe shortcuts, writing tests to detect dishonest or out-of-scope completions, and refining tasks based on QA feedback. The role requires strong Python and JavaScript/TypeScript skills, test design experience, familiarity with coding agents and GitHub PR/CI workflows, and English proficiency of B2+; the project is estimated at 20–25 hours per week during active phases and pays up to $50 per hour equivalent.
