Frontier AI labs are training models to do corporate work. Financial planning, deal structuring, compensation design, enterprise selling.
Mercor connects finance leaders, GTM operators, marketing professionals, HR executives, and business operators with the AI labs building these systems. Remote, paid project work starting at $50/hr.
Labs are building models that plan budgets and structure transactions. The people grading those models decide what the models learn to treat as correct.
Contribute from anywhere and choose projects that fit your schedule. Work independently without long-term commitments or fixed hours.
Hourly projects range from $50 to $100+/hour depending on expertise and project complexity. Task-based projects pay per completed deliverable, currently up to $1,000 per work product.
Every task you build gets used to score model output. Write the rubrics based on your business expertise.
Read model output the way you would read a first draft from someone on your team.
You build the deliverable yourself from mock source files. Then you write the rubric that determines how the next person's version gets scored.
Some projects ask you to break the model on purpose.
The last pass before work gets used. You check it against whoever would see it in a real organization.
Strategic decisions carry constraints that never make it into the documentation. A capital allocation logic applied by habit, or the reason a portfolio move stalled that nobody recorded.
Transactions turn on structure and on what a board will approve at a given moment. Accounting judgment shows up where the standards leave room, particularly when materiality is genuinely in question.
A $1M ACV enterprise deal involves a buying committee, a procurement cycle that adds months, a legal review that often kills the original structure, and a champion who changes jobs halfway through. Almost none of that is written down anywhere public.
Judging creative work means holding it against a brief and a business objective at the same time. That calibration is difficult to describe and harder to fake.
Roadmap decisions get made against incomplete evidence, competing internal priorities, and a governance process that moves at its own speed.
Compensation and employee relations decisions sit inside employment law that varies by jurisdiction. A change that clears Title VII and FLSA in the US can still trigger consultation obligations under EU labor directives.
Decisions move through organizations slowly and unevenly. Approval chains, vendor negotiations, and dependencies that surface late are hard to appreciate from outside a company.
Evaluate
Create
Test
Review
Every listing asks for five or more years in your function. Beyond that, the bar varies by project. Some ask for direct ownership at Fortune 500 scale, including CHRO-level people programs, board-level strategic recommendations, or F500 product lines. Others ask for deep craft fluency without requiring executive scope.
Both operating and advisory backgrounds work, and different projects want different things.
Our teams actively recruit from consulting and agencies such as McKinsey, BCG, Bain, Korn Ferry, and top-tier global creative agencies.
No. On most projects you are defining what excellent work looks like and explaining your reasoning in writing. Some projects ask you to produce the reference deliverable. Others ask you to score work someone else produced.
No AI background is required. Prior experience with AI training or evaluation work is a plus on some listings.
Compensation varies by project, function, and experience. Some projects pay hourly. Others pay a fixed rate per completed deliverable.