$65–90/hr · Mercor · Part time, 35 hours a week
A finance expert who evaluates AI models on financial reasoning and creates training data grounded in professional practice.
What you would do
- Assess AI-generated financial analyses, valuations, and investment recommendations against professional standards
- Create challenging, real-world finance scenarios with accurate solutions based on your experience
- Develop scoring rubrics and evaluation frameworks tailored to finance reasoning tasks
- Provide detailed written feedback on model correctness, judgment quality, and reasoning soundness
- Collaborate with other finance experts to standardize evaluation criteria
Who they want
- 8+ years working in investment banking, asset management, corporate finance, or financial advisory at recognized top firms
- Proven track record of career progression from analyst to associate to VP or director level
- Prior hands-on experience evaluating LLM outputs against structured criteria or rubrics
- Strong verbal and written communication for explaining financial reasoning clearly
- Ability to commit at least 35 hours per week on weekdays
What the interview asks about
1.Evaluating complex financial models
The interviewer needs to confirm you can assess whether an AI's financial model structure, assumptions, and calculations align with actual practice, since evaluation accuracy directly affects training data quality.
For example: “A model suggests that a high-growth tech company trading at 25x revenue should be valued using stable-growth DCF assumptions. What flaws would you identify in this reasoning, and how would you score it against a rubric?”
2.Task and solution design
Your ability to create realistic, graded problems with authoritative answers determines whether AI learns on problems that matter in professional finance.
For example: “Walk me through creating an M&A evaluation task where an AI must recommend which target a company should acquire. What would you specify about deal terms, synergies, and integration risk to make it solvable yet challenging?”
3.Rubric specificity and consistency
Clear, detailed evaluation frameworks prevent scorer drift and ensure consistent training data quality, which directly influences how well AI generalizes to real financial problems.
For example: “You're writing a rubric for scoring an AI's analysis of refinancing debt at higher rates to shorten maturity. What dimensions would you evaluate? How do you distinguish partially correct from fully correct?”
4.Identifying real-world failure modes
Models often succeed on paper but fail in practice; your job is to catch these gaps so AI learns realistic financial judgment, not just formula-driven answers.
For example: “A model correctly calculates that a specific hedging strategy minimizes portfolio variance. But in a market shock, the hedge becomes illiquid. How would you incorporate this real-world constraint into your evaluation?”
5.Working across domains with other specialists
Finance reasoning often touches tax, regulatory, and operational considerations; collaboration ensures evaluation standards remain coherent across interconnected areas.
For example: “You're harmonizing evaluation rubrics with a tax specialist and a compliance advisor. A model recommends a tax-optimal structure but mentions regulatory red flags. How would you define success when domain boundaries overlap?”
A task you may get
Evaluate a sample AI-generated investment memo recommending whether to buy or short a stock, assessing the quality of financial reasoning, data usage, and risk analysis against a professional standard.
How to prepare
- Review recent financial transactions or valuations you've worked on and be ready to explain your judgment calls
- Bring examples of financial models or analyses you've built or critiqued in professional roles
- Prepare a 3-5 minute overview of how you would grade an AI-generated financial recommendation
The facts
- Pay
- $65–90/hr
- Hours
- Part time, 35 hours a week
- Where
- Remote · United States
- Open to
- USA
- Field
- Finance
- Posted
- 7/10/2026
- Places left
- 10
We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.