Training Turk

Materials Science Expert

$60–120/hr · Mercor · Part time, 40 hours a week

A materials science expert who evaluates AI models on structure-property relationships and trains them on processing and service-condition realities.

What you would do

  • Create materials problems from systems you have personally processed and characterized
  • Score AI-generated explanations of how composition and microstructure drive material properties
  • Assess whether AI's proposed materials and processing windows are actually manufactureable
  • Grade AI's understanding of fatigue, corrosion, thermal cycling, and other service-condition limits
  • Provide feedback on the gap between theoretical material design and practical fabrication

Who they want

  • Graduate degree in materials science or related engineering field
  • Research or industrial experience where you personally processed and characterized materials
  • Clear written communication for explaining technical reasoning and material behavior
  • Comfort with ambiguous or incomplete instructions and willingness to flag unclear questions
  • Recent enough hands-on work that you can explain and justify your material choices

What the interview asks about

  1. 1.Structure-property linking

    AI must understand how processing creates microstructure and how microstructure drives mechanical and chemical properties; your judgment determines whether the model learns true materials science or just pattern-matching.

    For example: “You processed an aluminum alloy with two different heat treatments. The AI claims both achieve the same yield strength but differ in corrosion resistance. What characterization data would you need to see to score this explanation as correct or incomplete?”

  2. 2.Manufacturing feasibility

    Models often propose material specs that are theoretically optimal but impossible to manufacture reliably; your role is to catch these and teach AI what's real.

    For example: “An AI recommends a specific cooling rate during casting to achieve desired grain size. Based on your experience, this rate is unrealistic for industrial-scale production. How would you design a task and rubric that teaches AI about processing constraints?”

  3. 3.Service-condition failures

    Materials often meet initial design specs but fail under repeated thermal cycling, stress-corrosion cracking, or fatigue. AI needs to learn these modes.

    For example: “A model selects a nickel-based superalloy for an application based on high-temperature strength. You know the alloy is prone to oxidation at the operating temperature. How would you score and feedback on this recommendation?”

  4. 4.Characterization interpretation

    AI needs to understand what your experimental data actually reveals about material behavior; misinterpreting SEM, XRD, or mechanical test results is a critical error.

    For example: “You're writing a task where an AI must interpret microstructure images and mechanical property curves to infer processing history. What would constitute a correct vs. incomplete explanation?”

  5. 5.Ambiguity and incomplete specs

    Real materials problems often have missing or conflicting requirements; AI and evaluators need to flag gaps rather than make silent assumptions.

    For example: “You receive a task spec that asks for a high-strength, corrosion-resistant material suitable for marine environments but doesn't specify cost or processing scale. How would you flag this for clarification?”

A task you may get

Given a materials system you know well, write a design problem specifying composition, processing target, and property requirement. Then evaluate a sample AI response for technical soundness.

How to prepare

  • Identify a material system you have personally worked with and be ready to describe its processing, microstructure, and properties
  • Prepare one example where a reasonable design choice failed in real service conditions
  • Think about how you would teach someone new to materials science why theoretical predictions and practical results sometimes diverge

The facts

Pay
$60–120/hr
Hours
Part time, 40 hours a week
Where
Remote
Field
Other Engineering
Role type
Talent network
Posted
9/15/2026

We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.