We are recruiting experienced finance professionals for a company that tests how well leading AI models handle real finance work: valuation and modelling, credit analysis, investment memos, budgeting and variance analysis, treasury and risk. You bring the judgment you have built sending analyst notes back for buried assumptions, defending a model to an investment committee, or explaining a variance to a CFO. The AI output that judgment is needed to grade is supplied.

You will build tasks from your own practice, such as a three-statement or DCF model with a valuation summary, a credit memo with covenant analysis, an investment committee memo, a budget-versus-actual variance analysis, a hedging or liquidity proposal, or a portfolio attribution. You then run them through the AI and grade what comes back.

Some tasks are built around a handful of files. Larger scenarios take several days. The aim each time is a task a competent professional would get right and the AI currently gets wrong. This is not a traditional finance role. The work is long form and self-directed, and clear written reasoning matters as much as technical depth.

1. Design realistic finance tasks from your own day-to-day work, with the files and brief a colleague would need.
2. Run them through leading AI models and judge the results against the standard you would hold a colleague to.
3. Compare two AI answers to the same task and explain which is better and why.
4. Write grading rubrics: what a correct answer must contain, and why a response passes or fails.
5. Flag concrete failures with evidence, such as broken model logic, unsupported assumptions, misread financials, unit and scaling errors and missed risks.

1. 2+ years of hands-on experience in corporate or investment finance preferred. Less experience is considered where the practical work is real.
2. A bachelor's degree or higher in finance, economics, business or a quantitative field, or in progress.
3. Hands-on practitioner: you build and audit models yourself (three-statement, DCF, comps, credit or budget), at individual contributor level.
4. Clear, strong written English. You can explain why an answer is wrong, not only that it is.
5. Used AI tools such as ChatGPT or Claude at work, and can tell a sound answer from a plausible wrong one.
6. CFA, CPA, FRM or securities licences are welcome but not required. Practical expertise outweighs credentials.

Depth in at least one area is enough: equity or fixed income research, portfolio management or trading, corporate finance and FP&A, credit analysis, investment banking or M&A, treasury, market or credit risk, wealth management, or insurance and actuarial work. A working understanding of the others helps.

1. Pay: $40 to $125 per hour (USD), paid by PayPal on a regular cadence.
2. Contract role, fully remote. At least 10 hours a week, no weekly maximum. Full-time hours are available.
3. Flexible scheduling. Tasks have multi-day timers, so you can spread the work across the week.
4. Access to the AI tools is provided. You do not need a personal subscription.
5. Open to people based in the United States, Canada or the UK.

You start with a skills assessment of about 2 hours, which is unpaid. It tests your subject knowledge, careful reading, and how you judge an AI answer against a professional standard. All work must be your own. Submissions produced with AI tools are rejected, and this is the most common reason people do not pass.

If you pass, you complete onboarding, do a short training project, then move on to regular paid projects in your field. You are paid for every step you complete. Not every submission is accepted, and feedback and revision are a normal part of the process. Support comes through written guides, a dedicated Slack channel and office hours.

The company behind this works with leading AI labs and wants practising professionals, not AI training AI. If that suits you, apply and we will send the details.