Compare responses for Dutch corporate financial law documents

About OpenTrain

OpenTrain is a central job board for data-labeling and AI-training work, aggregating projects from many AI companies and labeling platforms so contributors can find this kind of work in a single place.

Creating an OpenTrain account is free and applying to projects typically takes only a few minutes.

• Discover AI-training jobs across many platforms in one place.
• Fast, simple applications with a free OpenTrain account.

AI training and why this work matters

AI systems learn from large collections of human-created examples. Tasks include evaluating and editing model outputs, comparing responses to expert (gold) answers, and flagging errors so models improve.

Evaluation work directly shapes model behavior: your comparisons help developers know where models fail, what kinds of mistakes recur, and how to train safer, more accurate systems.

• Work is fully remote and often lets you set your own hours.
• Contributions make models more accurate, safe, and useful.

The role

You will compare 80 pairs of responses: each pair is a model response and a gold (subject-matter expert) response to a Dutch corporate financial law query, with supporting documents provided.

For each pair you will produce a clear comparison and then deliver a short, aggregated report highlighting recurring issues and high-level insights across all 80 comparisons.

• Compare every one of the 80 model/gold pairs and document differences.
• Provide top-level insights summarizing common error patterns and improvement opportunities.

What you'll do day-to-day

Work through the 80 query-response pairs, reading the query, supporting documents, the gold response, and the model response for each item.

Produce concise, consistent comparisons that point out factual errors, missing elements, reasoning gaps, ambiguity, style or tone issues, and alignment with the gold answers.

• Create one comparison note per response pair describing how the model output differs from the gold.
• Tag or describe the type of issue (e.g., factual inaccuracy, omission, incorrect legal interpretation, unclear wording).
• Compile an aggregated summary of recurring mistakes and recommendations for improvement.

Requirements

This is an intermediate-level contract role. The project requires at least 20+ hours per week and is open to contributors worldwide.

Ideal fit: an English-speaking Dutch master’s student in law. Corporate financial law experience is a bonus but not critical, since many queries and documents are straightforward.

• Experience level: Intermediate.
• Time commitment: 20+ hours/week.
• Work type: Contractor, part-time, remote, worldwide.

Project details and tools

Data type is text: you will read Dutch legal documents and responses. Label type is PROMPT_RESPONSE_WRITING_SFT and the project uses OTHER labeling software (the platform will be specified when you start).

There are 80 total comparisons to perform and a separate top-level insights deliverable summarizing patterns across the set.

• Data type: TEXT.
• Labeling type: PROMPT_RESPONSE_WRITING_SFT.
• Labeling software: OTHER (details provided at onboarding).

Compensation, logistics, and how to apply

Pay is hourly at USD 40 per hour. This is a contract, part-time role with a minimum expected commitment of 20+ hours per week and is open to applicants worldwide.

To apply, create a free OpenTrain account and submit your application; applying typically takes only a few minutes. Additional onboarding and the exact tool link will be provided to selected contractors.

• Hourly rate: $40 USD per hour.
• Employment type: Contractor, Part-time.
• Geography: Worldwide applicants welcome.

Who should apply

Apply if you have law training (Dutch master’s students are a great fit), strong attention to detail, and can read Dutch legal text while working in English for documentation and deliverables.

Corporate financial law knowledge is helpful but not required — the crucial parts are careful comparison, consistent judgment, and clear written summaries of differences and trends.

• Ideal candidate: English-speaking Dutch law master’s student.
• Bonus: background in corporate financial law, though not mandatory.