Use your frontend craft to build hard, verifiable coding tasks from visual material. You turn screenshots, mockups, diagrams and screen recordings into multi-step tasks with objective success criteria, and back each one with a working reference implementation. Remote, project-based, and a public GitHub track record is the entry ticket.
What you will do
- Design multi-step coding tasks based on screenshots, UI mockups, diagrams or screen recordings of real workflows.
- Specify the code and artifacts each task involves and the behavior expected.
- Define objective success criteria, so anyone can verify a task was done correctly, not just plausibly.
- Write a working reference implementation for each task.
- Review tasks from other contributors for technical accuracy and verifiability.
- Flag tasks that turn out ambiguous or impossible to grade instead of submitting them.
What you need
- A public GitHub profile with substantial, verifiable contributions: commits, merged PRs or repositories you maintain. This is essential.
- Strong frontend engineering in React, TypeScript or similar, with experience shipping complex interfaces and interactions.
- The ability to read, change, debug and validate real application code.
- UI, graphics, SVG or Canvas experience is especially valuable.
- Product or UI designers with hands-on frontend work and public code contributions also fit.
- Strong written English and the ability to explain expected behavior and reproducible verification steps.
- A plus, not a requirement: data labeling, AI evaluation or rubric-based review experience.
How pay works
Pay is per approved task, about $640-800 per task. A task takes about 8 hours, which works out to the $80-100/hr equivalent listed above.
How applying works
- Apply and answer the screening questions.
- Complete the AI interview for this role.
- Your profile is reviewed.
- Onboarding to get you set up.