Data Science Expert

AI labs need analysis checked by people who have put models and experiments into production. A model can explain how an A/B test should be set up and still miss the leaked feature that makes a result look too good. Catching that takes a data scientist who has been burned by it.

This listing is a waitlist, not an open position. No project is attached to your application. Applying puts you in the network we draw from when a data science project opens. We ask for one thing: a short AI interview about your field, about 20 minutes, that you take once you have applied. We review applicants from time to time, check whether each profile fits the kind of projects we run, and reach out directly with the scope, hours and rate when there is a match. Timing depends on demand, so you may hear from us soon or only after a while, and we will not contact everyone on the list. You commit to nothing by applying.

Work varies by project. Depending on the project you might write analysis and modeling tasks with worked solutions, review other data scientists' work against a checklist, or grade AI output on statistics, experiment design and code. Hours, scope and rate are set per project and sent to you in writing before you start.

On a project you write realistic data science tasks and the answer an experienced practitioner would give, review other experts' work against an acceptance checklist, and evaluate AI answers for statistical errors, flawed methodology and code that runs but answers the wrong question. Reviewers check everything, and you are paid for approved work.

KEY RESPONSIBILITIES Write realistic analysis, modeling and experiment design tasks with worked solutions Review other experts' work against an acceptance checklist and explain what is wrong and why Evaluate AI-generated analysis and code for statistical validity and methodological errors Document your reasoning so a colleague could reproduce it Apply reviewer feedback and revise until the work is accepted Work from public or synthetic data only, with nothing confidential IDEAL QUALIFICATIONS Professional experience in data science, analytics or applied statistics Fluent in Python or R and SQL Able to explain a methodological choice in plain written English NICE TO HAVE Experience running or analyzing online experiments Machine learning in production A graduate degree in a quantitative field Prior work on an AI training or data platform WHAT SUCCESS LOOKS LIKE You apply and complete the short AI interview When a project matches your background we contact you with the scope, hours and rate You decide whether to take it, with no obligation either way Work you do is reviewed and you are paid for approved work.