Training Turk

Atomistic & Surface Modeling Experts (Computational Materials & Catalysis)

$84/hr · Mercor · Hourly, 40 hours a week

A computational materials expert who validates surface modeling and reaction prediction to train AI systems in heterogeneous catalysis and materials simulation.

What you would do

  • Develop, review, and adjudicate solutions involving electronic structure, surface modeling, adsorption, and reaction phenomena using first-principles and molecular methods
  • Create high-level research questions in computational materials science covering slab models, transition state discovery, and kinetic simulation
  • Evaluate AI-generated scientific outputs for computational correctness, physical validity, and mechanistic reasoning
  • Structure simulation setups, computational workflows, and results into well-organized training data that captures domain reasoning
  • Rate competing solutions against scientific criteria including accuracy, methodology rigor, and explanation quality

Who they want

  • PhD in materials science, physics, chemistry, chemical engineering, or closely related field with 3+ years post-PhD research experience
  • Hands-on expertise with first-principles codes (DFT, ab initio MD) and classical methods (MD, Monte Carlo) for materials and surface systems
  • Experience with standard computational software including VASP, Quantum ESPRESSO, CP2K, GPAW, LAMMPS, ASE, and pymatgen
  • Background in semiconductor materials or heterogeneous catalysis strongly preferred, with working knowledge of surface chemistry and microkinetics
  • Clear technical communication and ability to explain computational reasoning and material properties concisely in writing

Main skills

First principles calculations and dftSurface and interface modelingAdsorption and reaction energetics

What the interview asks about

  1. 1.DFT setup and convergence validation

    Correct setup of pseudopotentials, basis sets, and k-point sampling determines computational accuracy; AI must learn to spot invalid calculations.

    For example: “An AI sets up a DFT calculation of CO adsorption on a Ni(111) slab using GGA functional without dispersion correction and a 2x2x1 k-point mesh. What's missing or problematic, and how would you fix it?”

  2. 2.Surface model construction and reconstruction

    Choosing slab thickness, handling of boundary conditions, and accounting for surface relaxation or reconstruction are non-obvious; incorrect models give misleading results.

    For example: “You're modeling O2 dissociation on a reconstructed Fe3O4(001) surface. Why is using a simple bulk-truncated slab inadequate, and what additional considerations matter for this surface?”

  3. 3.Transition state searching and reaction barriers

    Predicting activation barriers requires not just finding saddle points, but validating that you've found the correct transition state with proper connections to reactants and products.

    For example: “An AI uses NEB to find what it claims is the rate-limiting transition state for a CO2 reduction step, reporting an activation barrier of 0.5 eV. What validations would you perform, and what could go wrong?”

  4. 4.Computational method selection and limitations

    Different simulation approaches (DFT vs MD, different functionals, implicit vs explicit solvation) have different accuracy/cost trade-offs that experts must choose wisely.

    For example: “You need to model how a catalytic poison adsorbs on a supported metal particle with 100+ atoms. Would DFT alone be feasible? How would you approach this problem and what approximations would you use?”

A task you may get

Design a research-level problem in surface modeling or catalytic reaction prediction, provide your complete solution using appropriate computational methods, and explain the key physical insights your calculation reveals.

How to prepare

  • Review recent papers from leading computational catalysis and materials groups to stay current with methodology and research directions
  • Practice running calculations with VASP or Quantum ESPRESSO and ensure fluency with convergence testing, functional selection, and output interpretation
  • Study transition state searching methods and understand the differences between NEB, CI-NEB, climbing image approaches, and dimer methods
  • Prepare two examples of published calculations where methodology limitations led to incorrect predictions, and discuss what went wrong

The facts

Pay
$84/hr
Hours
Hourly, 40 hours a week
Where
Remote · Remote — US based
Open to
USA
Field
Life, Physical, and Social Science
Posted
8/6/2026
Places left
3

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