Training Turk

Mercor listing

Multimodal Image Expert

$30/hr

The role in one line

You will judge AI‑generated image pairs for quality and prompt fidelity, providing concise written feedback.

Written by Training Turk from the public listing; it may be incomplete or out of date.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 two images side‑by‑side for each prompt
  • Assess alignment with the prompt, visual realism, and technical artifacts
  • Choose a preferred image or note equivalence/deficiency
  • Document a brief justification for the choice

Who they are looking for

  • Fluent English speaker with strong visual literacy
  • Ability to articulate observations in writing
  • No formal degree required, but a keen eye for detail
  • Available for hourly work after passing a $20 onboarding quiz

Skills this role asks for

visual assessmentprompt alignmentimage quality analysisdetail orientationwritten rationaleenglish fluency

What the interview is likely to probe

  1. 1.Prompt fidelity assessment

    The interviewer wants to confirm you can accurately map textual descriptions to visual outcomes, a core part of the rating task.

    Expect something like: “If a prompt describes a "sunset over a mountain lake," how would you evaluate an image that captures the colors but misplaces the horizon line?”

  2. 2.Anatomical accuracy detection

    They need to ensure you can spot subtle errors in human or animal forms, which affect overall quality scores.

    Expect something like: “When presented with two portraits, one has slightly distorted hand proportions—how do you decide which image better respects anatomical norms?”

  3. 3.Artifact identification

    Identifying rendering glitches is essential for consistent scoring across datasets.

    Expect something like: “Given an image with noticeable compression artifacts around text, how would you describe the impact on its overall rating?”

  4. 4.Written rationale clarity

    Clear explanations support downstream model training and auditability.

    Expect something like: “Write a 30‑word justification for preferring Image A over Image B when Image A has cleaner edges but slightly off‑color balance.”

Exercise you may get

You will be given a prompt and two generated images and asked to pick the better one while providing a brief written justification, similar to the onboarding quiz.

How to prepare

  • Practice evaluating AI‑generated images against prompts on public datasets
  • Write concise critiques focusing on prompt match, anatomy, and artifacts
  • Review examples of clear, structured feedback to emulate the expected style

Facts

Pay
$30/hr
Commitment
hourly
Work arrangement
remote · Remote
Domain
Arts & Design
Company
Moonlight
Posted
10/3/2026
Open slots
1839