The work
You will review large language model responses on corporate banking topics and judge whether the answers are accurate, well reasoned, and useful. You will also help create clear standards for evaluating banking judgments.
The work combines practical corporate banking knowledge with structured AI evaluation. You will collaborate with AI researchers and other banking experts on evaluation methods, training approaches, and performance benchmarks.
- Evaluate responses about syndicated and bilateral loans, leveraged finance, credit agreements, cash management, and trade finance.
- Create rubrics for leverage and coverage ratio analysis, debt capacity sizing, revolving credit facility structures, covenant compliance, and credit ratings.
- Apply consistent standards when reviewing model accuracy, reasoning, and usefulness.
- Use knowledge of corporate credit, capital structures, loan syndication, and relationship coverage for large and multinational companies.
What it pays and takes
This is a fully remote, flexible freelance contractor role. It is listed as entry level and is expected to run for about one month, with a possible extension based on performance and project needs.
- Pay: $100 per hour.
- Schedule: 20+ hours per week.
- Location: Open worldwide.
- Language: Excellent written English is required.
- Experience: At least two years of corporate banking experience, including work at a large commercial or investment bank.
- Relevant backgrounds include corporate coverage, credit risk, loan syndications, leveraged finance, and transaction banking.
- Required skills include critical thinking, problem solving, and the ability to explain complex banking judgments clearly.
- Prior AI experience is not required.
- Helpful but optional: CFA, FRM, CA, CPA, or an MBA in Finance.
How it works
You apply on OpenTrain with your resume and then complete the application on the hiring site. OpenTrain is where you find this role and begin the application process.
About AI training work
AI training is the human work behind systems that generate and understand text. Contributors review examples, rate model answers, and explain what makes an output accurate or useful, while experienced professionals bring the subject knowledge needed for specialized evaluations.