Compensation: $80 - $160/ hr
Overview
Huzzle Labs is seeking experienced data scientists to review enterprise slide content and rate it for substance and style. You will read business presentations such as analytics readouts, experiment results and ML project updates, and judge how rigorous, insightful and clearly communicated they are. Your judgement builds the reference data used to evaluate frontier AI systems.
The engagement starts with a short screening task in which you rate enterprise slides for content and style. Depending on your screener performance, you may be invited to take part in ongoing campaign tasks.
This is an expert judgement role focused on careful reading, precise rating and consistent documentation.
Key Responsibilities
- Rate slides: Score enterprise slides and decks for content (analytical rigor, insight, logic, fit for the audience) and style (structure, clarity, data visualization, polish).
- Judge analytical substance: Assess whether methods, metrics and conclusions are sound and whether the results are communicated honestly.
- Explain your ratings: Write short, precise rationales that a peer reviewer could follow.
- Apply guidelines consistently: Follow the rating rubric, flag ambiguous cases and help refine the criteria.
- Deliver on time: Work in batches and meet turnaround targets.
Ideal Qualifications
- Experience as a data scientist at a top technology company.
- An undergraduate degree from a well-known English-speaking university: top 100 in the US News Best Global Universities rankings (English-speaking countries only), or top 300 in the US News US college rankings.
- At least 1 year of professional work experience, with part of that employment within the past 7 years.
- Fluent written English and a strong sense of what good business communication looks like.
- Careful and detail-oriented: consistent ratings and clear written rationales.
- Nice to have: experience building or reviewing executive-ready presentations; prior annotation, rating or QA experience.
About Huzzle Labs
Huzzle Labs partners with leading AI organisations to evaluate and improve frontier models through deep human expertise. Contributors assess advanced AI systems in their specialist fields, receive competitive compensation, and help define the standards used to measure model performance.
