Computational Chemistry AI Evaluator at Mercor

Computational Chemistry AI Evaluator | $70–100/hr | Worldwide Remote

Join a cutting-edge project building large-scale evaluation benchmarks for advanced AI reasoning in scientific and engineering domains. As a task designer, you'll create graduate-level computational problems that challenge AI systems to use real scientific software tools — from running simulations and interpreting outputs to designing experimental strategies and recovering hidden information from data.

This is not a typical annotation or labeling role. You'll be crafting original, research-grade problems grounded in real scientific workflows, calibrating them against frontier AI models, and iterating until the difficulty is precisely right.

What You'll Do

• Design sophisticated computational problems requiring domain-specific scientific software libraries
• Create tasks that test multi-step scientific workflows and strategic experimental reasoning
• Participate in a calibration loop — testing problems against state-of-the-art AI models and refining difficulty
• Write problem setups, oracle functions, and solution validators in Python

Domain Focus: Computational Chemistry & Electronic Structure

We're seeking experts with deep, hands-on experience using PySCF for quantum chemistry calculations, including Hartree-Fock, DFT, TDDFT, CASSCF, and post-HF methods. Ideal candidates can design problems around excited-state analysis, orbital diagnostics, method selection for complex electronic structures, and interpreting computational artifacts from method limitations. Experience with other specialized software in this domain is also considered.

Requirements

• Graduate-level training in a relevant STEM domain (MS, PhD, or equivalent research experience)
• Demonstrated proficiency with at least one listed scientific software library (evidenced by publications, open-source contributions, or professional work)
• Strong Python programming skills
• Ability to work independently and iterate based on calibration feedback
• Comfortable in a Linux/terminal environment with remote compute sandboxes
• Available for at least 15–20 hours per week

Nice to Have

• Experience across multiple scientific domains or tools
• Familiarity with benchmark or evaluation design
• Background in scientific pedagogy or problem-set design
• Experience with computational reproducibility and containerized environments

Strong candidates will think like puzzle designers — constructing problems where difficulty stems from reasoning strategy, not brute computation, and where surface-level pattern matching won't suffice.