Statistical Physics and Thermodynamics Physicist (PhD)

The role in one line

You create, solve, and evaluate research-level physics problems in your specialized subdomain to help benchmark whether AI models can perform genuine physics research reasoning.

Written by Training Turk from the public listing; it may be incomplete or out of date.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

What the interview is likely to probe

1. 1.Problem design at research frontier

   Distinguishing between textbook-solvable problems and genuine research questions reveals whether you understand the limits of standard techniques and can identify open questions in your field.

   Expect something like: “Design a physics problem on the Nishimori line that requires understanding gauge symmetry and quenched disorder averaging. Explain why this problem tests research reasoning that textbooks don't cover.”

2. 2.Exact lattice method execution

   Exact methods require meticulous bookkeeping and deep understanding of symmetries; competence here distinguishes researchers who can hand-verify answers from those who rely on numerical approximations.

   Expect something like: “Show how you'd use Kramers-Wannier duality to derive the partition function for a specific square-lattice model with twisted boundary conditions. What symmetries do you exploit and why?”

3. 3.Replica trick and analytic continuation

   The replica trick appears throughout statistical physics but requires care with analytic continuation; mishandling it is a common source of errors, so evaluating your mastery is essential.

   Expect something like: “You compute entanglement entropy using the replica trick, obtaining an expression in integer replicas n. Describe how you'd continue this expression analytically to n approaching 1, and what pitfalls might arise.”

4. 4.Conformal field theory and fusion rules

   CFT requires comfort with abstract algebraic structures like Virasoro algebras and fusion rules; errors here propagate silently unless caught by someone with deep technical knowledge.

   Expect something like: “For the Ising minimal model, explain how the operator product expansion and crossing symmetry constrain the three-point correlation functions, and how you'd verify a proposed set of conformal blocks.”

5. 5.Asymptotic analysis and logarithmic corrections

   Asymptotic expansions often hide subtle logarithmic corrections that dramatically affect physical predictions; catching or correctly handling these distinguishes careful work from sloppy derivations.

   Expect something like: “A range expansion model with heavy-tailed dispersal suggests an expansion with leading term A log(t). Describe how you'd determine the subleading terms and verify your asymptotic series is consistent.”

Exercise you may get

Create a research-level problem in your specialty that requires mastery of your field's key methods, solve it completely with written explanation, then describe what the correct solution tells us about physical behavior that a textbook problem would miss.

How to prepare

• Review 2-3 of your published papers and identify the most technically demanding derivations, then practice explaining them clearly to a colleague in a slightly different subfield
• Select one of your papers' key results and trace through every mathematical step, documenting exactly which specialized technique is essential at each stage
• Draft a problem mirroring a research concept but adapted to avoid direct copying; solve and evaluate whether it requires specialized domain expertise.
• Collect examples from recent arXiv papers in your subfield where authors clearly work through difficult derivations and study how they structure mathematical reasoning in writing

Facts

Pay

$80–110/hr

Commitment

hourly

Hours

10 per week

Work arrangement

remote · Remote

Domain

Life, Physical, and Social Science

Posted

9/25/2026

Open slots

3