Training Turk

Mercor listing

Data Science Expert

$100–150/hr

The role in one line

A data science expert who evaluates AI-generated analyses, spreadsheets, and reports for methodological rigor and professional quality.

Written by Training Turk from the public listing; it may be incomplete or out of date. Read the full posting on Mercor.

What you would do

  • Review AI-generated data science outputs including presentations, dashboards, and analytical documents
  • Assess statistical methodology, formula correctness, and validity of conclusions
  • Evaluate data visualizations for accuracy, clarity, and suitability to the audience
  • Identify computational errors, unsupported claims, and gaps in reasoning
  • Provide clear, detailed written feedback explaining what makes an analysis professional and trustworthy

Who they are looking for

  • 5+ years of professional data science work at a recognized top firm in US, UK, Canada, Australia, or New Zealand
  • Native or professional fluency in English for clear written communication
  • High proficiency in Microsoft Office and Google Workspace, especially PowerPoint, Excel, and Docs
  • Master's degree or higher from a reputable institution is preferred
  • Ability to evaluate quality across diverse data science tasks and domains

Skills this role asks for

statistical methodologydata visualization designspreadsheet analysispresentation accuracyfactual error detectionanalytical reasoningbusiness intelligencedata interpretationquality assessmentdomain expertise

What the interview is likely to probe

  1. 1.Statistical methodology evaluation

    AI often produces outputs that look polished but rest on flawed statistical foundations; your judgment determines whether the model learns rigor or just surface credibility.

    Expect something like: “An AI creates a regression model predicting quarterly revenue with an R-squared of 0.94, but uses 15 predictors on 20 data points. What are the primary red flags you'd identify, and how would you explain them to a non-technical stakeholder?”

  2. 2.Visualization accuracy and clarity

    Misleading charts or incomplete legends can undermine even correct analysis; AI must learn to visualize data honestly and readably.

    Expect something like: “You evaluate an AI dashboard with a stacked bar chart showing revenue by region. It lacks a legend, uses similar colors, and omits one quarter. How do you score this, and what feedback do you give?”

  3. 3.Computational correctness

    Rounding errors, formula mistakes, or copy-paste bugs destroy credibility; your spot-checking teaches AI to validate its own math.

    Expect something like: “A spreadsheet calculates customer lifetime value by multiplying average order value, purchase frequency, and retention rate. You spot-check three calculations and find errors in two. How would you assess the severity and what would you recommend?”

  4. 4.Business context fit

    Technically correct analyses that don't answer the business question are useless; AI needs to learn to translate analytical findings into actionable insights.

    Expect something like: “A stakeholder asks whether to expand into a new market. The AI provides a 20-page analysis of market size and growth but omits unit economics, competitive positioning, and execution risk. How do you evaluate this?”

  5. 5.Assumption clarity and disclosure

    Professional data science requires explicit discussion of assumptions, limitations, and confidence intervals; AI tends to paper over these.

    Expect something like: “An AI predicts customer churn at 92% accuracy but omits class imbalance disclosure, confusion matrix, and segment-level performance. What's required for professional-grade output?”

Exercise you may get

Evaluate a sample AI-generated data analysis or business report. Identify at least three distinct issues spanning methodology, visualization, and accuracy. Write structured feedback explaining each issue and how to fix it.

How to prepare

  • Review a recent analysis you've critiqued at work and prepare to explain what separated acceptable from excellent work
  • Identify one dataset or business question you've tackled and be ready to describe the analytical approach and pitfalls to avoid
  • Think about the difference between statistically correct and business-useful analyses in your experience

Facts

Pay
$100–150/hr
Commitment
hourly
Hours
40 per week
Work arrangement
remote · Remote
Domain
Data Analysis
Company
Dorado
Posted
9/21/2026
Open slots
82