Data Scientist, Part-Time ML Task Authoring - Remote (United States)

OpenTrain AI; United States; Remote; USD 83,000 - 165,000; Part time

Benefits offered by this job

Fully remote

Flexible schedule

Independent contractor

Job summary

OpenTrain AI is seeking experienced data scientists to craft and evaluate challenging tasks for coding agents. You’ll design data science problems, assemble data and environments, and write reference solutions with tests and scoring rules.

Work remotely on your own schedule, paid per accepted task. You’ll prepare datasets, dependencies, and run requirements; verify reference implementations; set up evaluation tests; and guide coding agents through debugging and verification steps.

Qualifications

• Fluent in English.
• Strong Python skills for testing and debugging unfamiliar code.
• Background in data science, machine learning, quantitative modeling, or research competitions.

Responsibilities

• Prepare the execution setup, including the required data, dependencies, and run requirements.
• Write and verify reference implementations for each data science task.
• Set up tests and scoring rules to distinguish complete, correct work from partial or flawed approaches.
• Have coding agents attempt the tasks, trace failures, and record the checks used.

Skills

Python

Testing & debugging

Data science background

English fluency

Tools

CodaLab

Codabench

Kaggle

Numerai

Zindi

AIcrowd

DataSource.ai

DrivenData

Job description

• $60–$120 per hour equivalent, paid per accepted task
• About 15 hours a week, you choose when

Create challenging data science tasks and check how coding agents handle them, on your own schedule.

AI companies need competitive data scientists to design challenging tasks that assess data science coding. Set up repeatable execution environments, write and verify reference solutions, add tests and scoring rules, and have coding agents attempt the tasks. Trace failures, find implementation mistakes and weak assumptions, and record the task requirements, reasoning behind the solution, and checks used to verify it. This is that work, done remotely around your schedule.

What you'll do

• Prepare the execution setup, including the required data, dependencies, and run requirements
• Write and verify reference implementations for each data science task
• Set up tests and scoring rules to distinguish complete, correct work from partial or flawed approaches
• Have coding agents attempt the tasks, trace failures, and record the checks used

Why this is worth your time

• Earn $60–$120 an hour equivalent, paid per accepted task
• Use your data science and Python skills to build and verify real evaluation tasks
• Choose your own hours, about 15 a week
• Fully remote and flexible, no fixed shifts, no commute

Who this is for

• Fluent in English
• Have experience in competitions including CodaLab/Codabench, DataSource.ai, Numerai, AIcrowd, Zindi, DrivenData, or Kaggle
• Strong Python skills for testing and debugging unfamiliar code, together with a background in data science, machine learning, quantitative modeling, or research competitions
• No experience in AI training work needed.

Pay and details

• $60–$120 per hour equivalent; pay is per accepted task, as an independent contractor
• Part-time, about 15 hours a week; fully remote, United States

About OpenTrain AI

OpenTrain AI connects experienced professionals with paid, part-time AI training work: shaping and improving the AI systems that companies are building right now.