Competitive Data Scientist (Kaggle)

You will create and validate difficult data science tasks used to assess coding agents and data science ability. The work combines modeling, Python programming, debugging, testing, and careful review of whether solutions are correct.

• Build reproducible environments with the needed data, dependencies, and execution requirements.
• Implement verified reference solutions that show how each task should be completed.
• Create tests and evaluation criteria that separate correct solutions from incomplete or flawed approaches.
• Run coding agents against tasks, investigate failures, and improve task designs based on what you observe.
• Find implementation errors, flawed assumptions, data leakage, and other issues that affect correctness.
• Document task requirements, solution reasoning, and verification steps clearly.

This is a remote contractor role with a flexible schedule. The listed experience level is entry level, and the work is open to candidates in the countries supported for this project.

• Pay: $60 to $120 per hour equivalent, paid per task that meets the project specifications.
• Time: Approximately 15 hours per week, with fewer than 20 hours per week expected.
• Schedule: Choose the hours and days you work.
• Language: Fluent in English.
• Experience: Background in data science, machine learning, quantitative modeling, or research competitions.
• Competition experience with Kaggle, DrivenData, Zindi, AIcrowd, Numerai, DataSource.ai, or CodaLab/Codabench.
• Strong problem-solving skills and a clear understanding of the methods used in your solutions.
• Programming ability to implement, test, and verify solutions independently.
• Strong Python skills, including testing, debugging, and working with unfamiliar code.
• Published competition solutions, a strong competitive data science profile, or a notable competition rank can help demonstrate practical expertise.
• Applied mathematics, machine learning engineering, or quantitative research experience is relevant when paired with hands-on implementation and verification skills.

You apply here in about two minutes and finish on the hiring platform. The next step is a short AI interview (technical roles include a coding challenge). Passing it is how you qualify for paid tasks, so take it soon after you are invited.

AI training is the human work behind artificial intelligence, including preparing examples, testing model outputs, and reviewing whether computer-generated answers or code are correct. People with practical technical expertise help create reliable evaluations that improve how AI systems perform.