What the work actually involves

You are not managing money here — you are encoding the judgment you use to manage it. A typical task starts from your own mandate or book: write an investment case or client scenario with the thesis stated, the constraints named (liquidity, concentration limits, drawdown tolerance, time horizon), and the risk budget made explicit. Then you write the reference answer the model's attempt is graded against. Other tasks run the other direction — a model has already produced a portfolio, an allocation, or a tax strategy, and you grade it. Are the factor exposures intended or incidental? Is that the right benchmark? Does the withdrawal sequencing survive a bad first decade? Does the attribution actually explain the return, or just decompose it plausibly?

The highest-value output is usually the third category: flagging recommendations that are defensible on paper and would still fail a real client. A model rarely produces something obviously wrong. It produces something that ignores the 30% single-stock concentration the client won't sell for tax reasons, or rebalances a portfolio that has a lockup, or sequences withdrawals in an order that works pre-tax and not after.

What the platform screens for

Mercor verifies background and runs a roughly 20-minute AI interview. The screen is looking for whether you personally made allocation or portfolio calls, recently enough to reconstruct the tradeoffs — not whether the firm on your resume is well known. Expect follow-ups that push past the headline decision into why the alternative was rejected. Written clarity matters as much as the underlying judgment, because most deliverables are explanations rather than numbers. Comfort with ambiguous instructions and a willingness to say "this task is underspecified, here's what's missing" is explicitly part of the profile.

Logistics and honest expectations

  • Standing listing, not an opening. It stays live whether or not a project is running. You will not receive an accept or reject decision on this application.
  • Pool-based matching. After verification you sit in the asset and wealth management pool; specific project listings name the rate, hours, and client, and the hiring decision happens there.
  • Timing is genuinely variable. A match can come within a week or take several months, depending on what clients are running.
  • Remote and asynchronous, project-based hours rather than a fixed weekly commitment. Joining every network you qualify for improves match odds, and you only sign up once.