Overview
Remote AI training role designing and verifying competitive data science and machine learning tasks, and evaluating how AI coding agents perform on them.
What you'll do
- Design challenging data science and machine learning tasks inspired by real-world problems and competitive data science scenarios.
- Build reproducible execution environments, prepare datasets, configure dependencies, and implement verified reference solutions.
- Develop automated tests and evaluation criteria that distinguish correct solutions from incomplete, incorrect, or unreliable approaches.
- Use coding agents to execute assigned tasks, investigate their failures, and evaluate their problem-solving capabilities.
- Identify implementation errors, flawed assumptions, data leakage, statistical inconsistencies, and other issues affecting solution correctness.
- Apply competitive data science techniques and mathematical reasoning to validate solutions.
- Ensure tasks have clear requirements, reliable expected outputs, and reproducible verification procedures.
- Refine task difficulty and evaluation methods based on observed coding-agent performance.
- Document task specifications, solution methodologies, technical reasoning, and verification steps.
- Review and improve task quality to ensure scientific and computational accuracy.
What you'll need
- Demonstrated experience in competitive data science, machine learning, quantitative modeling, or applied mathematics.
- Kaggle profile with at least Expert ranking.
- Ability to provide a verifiable Kaggle profile demonstrating competition achievements and ranking.
- Strong understanding of machine learning, statistical modeling, data analysis, and quantitative problem-solving.
- Practical programming ability, particularly in Python.
- Ability to independently implement, test, debug, and verify data science solutions.
- Experience working with datasets, computational environments, and relevant machine learning libraries.
- Strong analytical reasoning and the ability to explain technical decisions.
- Ability to identify data leakage, implementation flaws, and methodological errors.
- Excellent written communication and technical documentation skills.
- Professional English communication skills.
Nice to have
- Master or Grandmaster ranking on Kaggle (strongly valued).
Details
- Location: Remote, within US, Canada, Australia, New Zealand, or UK.
- Workload: approximately 15 hours per week.
- Schedule: flexible; choose your working days and hours, including weekends.
- AI training experience is not required.
Read the full description and apply on the company’s own careers page.