Expert Rubric Task Author Needed $30 per task

Rubric Task Author — Project Bulls Eye (Contract, Paid Per Task)
About the Role

AI models are already being used for real professional work — writing reports, answering questions, building analyses. But they're not reliable enough yet: they still get things wrong in ways that matter, and right now nobody can consistently find out where.

We're looking for domain experts to help find those breaking points. You'll build realistic, professional-grade deliverables in your area of expertise, then build the rubric that grades them — creating prompts hard enough that today's best AI models still get wrong, but realistic enough that a working professional would genuinely need them answered correctly.

This work spans 67 occupations across 16 domain groups, including Healthcare, Finance & Accounting, Legal & Compliance, Engineering, Supply Chain & Procurement, and Business Operations.

What You'll Do
Design a realistic professional task from your field — the kind of question, report, or analysis a working professional would actually be asked to produce.
Write the prompt itself: the realistic, hard question or scenario a professional would need answered.
Build the deliverable yourself, to the standard a qualified professional would expect.
Build a rubric that precisely and objectively grades that deliverable — capturing exactly what separates a correct, professional-quality answer from a flawed one.
Stress-test the task against current AI models to confirm it's genuinely hard — the model should get it wrong or fall short in a meaningful way.
Pass the project assessment required to qualify for task work.
Submit your task for review and validation.
What We're Looking For
Real professional experience in one or more of the 16 domain groups (Healthcare, Finance & Accounting, Legal & Compliance, Engineering, Supply Chain & Procurement, Business Ops, or similar).
Deep enough expertise to know what a genuinely hard, realistic professional scenario looks like in your field — not a textbook question, but something with the ambiguity, edge cases, or judgment calls real work involves.
Ability to write clearly and define objective, defensible grading criteria — you're not just producing an answer, you're specifying what "correct" means.
Comfort working independently through a structured, guided workflow.
Requirements to Get Started
Complete and pass the Project Bulls Eye project assessment to qualify for task work.
For each task, write an original prompt and build out the full deliverable + rubric as described above.
Pay Structure
$30 per task, paid per task rather than by the hour.
Payment is issued only after your task is validated and passes review — meaning it clears the required quality checks and is confirmed to meet the project's standards before it's accepted.
Tasks that don't pass validation are not eligible for payment; you're welcome to revise and resubmit where applicable.
Engagement Details
Contract / freelance, task-based.
Work through the guided workflow in the Project Bulls Eye learning hub, then pass the project assessment before submitting your first task.
Best suited for professionals who want flexible, self-directed contract work tied to their existing domain expertise.

To apply, tell us which of the 16 domain groups and occupations best match your background, and briefly describe a real, hard professional scenario from your field that you think would stump an AI model today.