AI Data Science Domain Expert

$100–200/hrrate as listed

micro1 recruits practicing data scientists to write, review, and grade tasks that test how well AI models handle real analytical work — from experiment design to messy dataset diagnosis. Engagements are contract, remote, and paid hourly, with observed rates in the $100–200/hr range depending on depth and specialization.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work involves

You build and judge evaluation material for frontier models on data science reasoning. In practice that means authoring problems a competent model should plausibly fail — a leakage-riddled feature pipeline, an A/B test with a peeking problem, a time-series split done wrong, a metric choice that flatters a useless classifier — then writing the reference solution and the rubric that separates a defensible answer from a fluent one.

Day-to-day tasks typically include:

Writing prompts grounded in real analytical situations, with the data, code, or output a working analyst would actually see
Grading multiple model responses against a rubric and explaining, in writing, exactly where reasoning breaks
Reviewing other contributors' items for correctness, ambiguity, and unintended giveaways
Occasional live comparison work: ranking outputs side by side and justifying the ordering

What micro1 screens for

The intake is an AI-led interview followed by task-based work samples. Screening looks for whether you have shipped analysis that someone depended on — not coursework, not tutorials. Expect follow-up questions that push past your first answer: which estimator, why that assumption, what you'd do if the residuals misbehaved. Reviewers also weigh writing quality, because a rubric that cannot be applied consistently by a second grader is worthless.

Statistical honesty matters more than tool breadth. Contributors who flag uncertainty, distinguish correlation from claim, and refuse to grade outside their competence tend to get more work.

Logistics

Fully remote and largely asynchronous, with tasks pulled from a queue against deadlines rather than fixed shifts. Most contributors commit 10–20 hours weekly, though volume fluctuates with project cycles and some engagements ramp sharply for short periods. Pay is hourly under an independent contractor agreement; the $100–200/hr band reflects observed rates and is not a guarantee for any given project or region.

What they look for

critical thinking
analytical reasoning
attention to detail
quality assurance
written communication
technica documentation
prompt authoring & refinement
AI Output Evaluation
Professional Writing
Report Writing
Business Communication
Problem Solving
Content Review
Data Interpretation
Logical Resoning
Professional Editing
Independent Research
Data Annotation
Rubric-Based Evaluation
Content Evaluation
AI Model Evaluation
Fact Checking
RLHF
Technical Editing
Prompt Engineering

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Have you authored and delivered an analysis—not coursework or a tutorial—to a stakeholder or product team who took a decision or action based on your findings?
Shipped analysis experience · required
Do you have hands-on experience identifying and explaining statistical errors or risks in analyses (e.g., confounding, leakage, violated assumptions, selection bias)?
Statistical reasoning depth · required
Can you write clear, application-ready rubrics or grading guidelines that a second person unfamiliar with the problem could apply consistently?
Writing clarity · required
Are you able to commit to a minimum of 10 hours weekly over the next 3 months, with flexibility to scale toward 20 hours in weeks when project volume is high?
Availability window · preferred

0 of 4 answered

Included with this listing

Interview prep — 12 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Whether the candidate has hands-on experience authoring analyses that moved a decision, not just exploring data in isolation or running canned pipelines.
A strong answer: A concrete example with stakeholder, original question, the analytical step that mattered most, and evidence the answer was used (a decision made, a feature shipped, a budget moved)—not a tutorial or interview project.
Where people lose it: Naming a project but vaguely describing it, or describing process (cleaned data, ran models) without saying what question was answered or who relied on it.