Competitive Data Scientist (Kaggle)

Data science competitors design benchmark-style tasks that push coding agents beyond plausible-looking answers. Every task needs its own runnable environment, a checked reference solution and dependable scoring. Contributors run agents on their tasks, study where the agents fail, look for flaws such as data leakage, and refine the task and its documentation.

What you would do

• Build reproducible environments with data, dependencies and reference solutions
• Write tests that tell correct solutions apart from incomplete or flawed ones
• Run coding agents on tasks and investigate their failures
• Document the requirements, the logic of the solution and how it was verified

What the listing asks for

• Proven background in data science, machine learning, quantitative modeling or research contests such as Kaggle
• Strong problem solving and a clear grasp of the methods you use
• Enough programming skill to build and check your own solutions
• Preferred: strong Python, including testing, debugging and unfamiliar code

Suits people who have competed in data science contests, want flexible hours including weekends, and enjoy debugging how coding agents behave.

Listed pay: $60 to $120 per hour. Where: Remote, worldwide. Time: About 15 hours a week, flexible schedule. Contract work for micro1, a company that hires experts for AI training projects.

Applications go through micro1's own site. AI Trainer Guide is an independent guide to AI training work, not micro1 or its recruiter. The apply link on our page is a referral link: micro1 may pay us if it hires you, at no cost to you.