Description
Mindrift is seeking a project-based Software Engineering Evaluation Specialist for Terminal Bench to design realistic, reproducible coding tasks that test and improve AI coding agents. The specialist creates broken software environments, Docker configurations, pytest verification tests, Jira-style instructions, and reference solutions; calibrates task difficulty, incorporates QA feedback, and may review other authors' tasks. The role requires at least three years of production backend software development, strong Python and pytest skills, Docker and Linux/Bash expertise, experience with AI coding agents, and B2+ written English. Contributions are flexible and paid at rates up to $35 per hour, but this is not permanent employment.
