Physical Sciences Expert - Terminal Bench

Weekday AI; India; Remote; INR 684,000 - 951,000; Part time

Job summary

Weekday AI is seeking highly qualified Physical Sciences Experts to develop terminal-based benchmarking tasks for AI evaluation. You will translate workflows from physics, chemistry, materials science, astronomy, and computational science into reproducible, rigorous benchmark environments.

Candidates should hold a PhD or equivalent, with strong programming skills (Python, C/C++, Julia) and experience in numerical modeling, Linux, and reproducible research practices.

Qualifications

• PhD in Engineering or relevant physical-science discipline or equivalent experience.
• Strong expertise in at least one core domain (physics, chemistry, materials science, astronomy, cosmology, computational science).
• Proficiency in Python, C/C++, Julia, Bash, or other scientific computing languages.
• Experience with numerical methods, simulations, modeling, or data analysis.
• Familiarity with Linux and terminal-based environments.
• Ability to independently develop, debug, and validate workflows.
• Strong understanding of units, precision, reproducibility.

Responsibilities

• Design challenging, multi-step terminal-based tasks based on authentic physics research and workflows.
• Build reproducible computational environments with scientific software and dependencies.
• Create datasets, structures, parameters, and model configurations.
• Implement reliable solutions using Python, C/C++, Julia, Bash or domain tools.
• Develop tests to validate numerical accuracy and physical consistency.
• Design simulations, modeling, data fitting, optimization, and visualization tasks.
• Define units, tolerances, and boundary conditions for reproducibility.
• Ensure tasks run reliably without undocumented dependencies.
• Troubleshoot dependency conflicts and performance issues.
• Document scientific methods, assumptions, and limitations.

Skills

Python

C/C++

Julia

Bash

Linux

Numerical methods

Scientific computing

Terminal-based development

Problem solving

Education

PhD in Engineering or physical-science

Postdoctoral experience or equivalent

Tools

OpenMM

ASE

RDKit

Psi4

LAMMPS

GROMACS

Docker

Conda

Git

Job description

This role is for one of our clients

Role Details

• Employment Type: Contract
• Location: Bangladesh, India, Pakistan, Indonesia, Egypt, Ghana, Nigeria, Turkey, Brazil, and Colombia
• Contract Duration: 5 weeks
• Start Date: Immediate
• Work Commitment: 40 hours per week, with at least 4 hours of daily overlap with Pacific Standard Time (PST)
• Compensation: $300 per approved task
• Average Task Completion Time: 6–8 hours
• Contract Benefits: No medical benefits or paid leave

About the Role

We are seeking highly qualified Physical Sciences Experts to develop realistic, terminal-based scientific computing tasks for AI benchmarking and evaluation. The role involves translating authentic workflows from physics, chemistry, materials science, astronomy, and computational science into reproducible, technically rigorous benchmark environments.

You will develop scientific inputs, computational models, executable reference solutions, automated tests, and objective grading criteria. Your work will help assess whether AI agents can reason through complex scientific problems, use command-line tools effectively, debug calculations, execute simulations, and generate accurate, reproducible scientific outputs.

The ideal candidate combines advanced academic or research expertise in a relevant physical-science discipline with strong programming skills and practical experience in computational modeling, numerical analysis, and scientific data processing.

Core Scientific Domains

Candidates should have expertise in at least one of the following areas:

• Physics and computational physics
• Chemistry and computational chemistry
• Materials science and materials modeling
• Astronomy and cosmology
• Thermodynamics
• Quantum mechanics
• Statistical mechanics
• Numerical and computational science

Key Responsibilities

• Scientific Task Design: Develop challenging, multi-step terminal-based tasks based on authentic physical-science research and computational workflows.
• Environment Development: Build self-contained, reproducible computational environments with appropriate scientific software, dependencies, and configurations.
• Scientific Input Preparation: Create datasets, molecular structures, simulation parameters, experimental data, and model configurations.
• Reference Solution Development: Implement reliable scientific solutions using Python, Bash, C/C++, Julia, or other relevant programming languages and domain-specific tools.
• Automated Testing: Develop tests to validate numerical accuracy, physical consistency, convergence, expected output structures, and task completion.
• Computational Modeling: Design tasks involving simulations, numerical modeling, data fitting, optimization, spectroscopy, molecular analysis, and scientific visualization.
• Scientific Validation: Define appropriate units, numerical tolerances, boundary conditions, physical constraints, and expected scientific behavior.
• Reproducibility: Ensure tasks execute reliably in controlled environments without requiring runtime downloads or undocumented dependencies.
• Debugging and Optimization: Troubleshoot dependency conflicts, numerical precision issues, solver instability, computational performance, and file-format problems.
• Documentation: Clearly communicate scientific assumptions, computational methods, expected outputs, edge cases, and limitations to reviewers.

Required Qualifications

• A PhD in Engineering or a relevant physical-science discipline, or postdoctoral experience or equivalent advanced technical experience in a relevant field.
• Strong expertise in at least one core domain, such as physics, chemistry, materials science, astronomy, cosmology, or computational science.
• Proficiency in Python, C/C++, Julia, Bash, or another programming language used for scientific computing.
• Experience with numerical methods, scientific simulations, computational modeling, or quantitative data analysis.
• Familiarity with Linux and terminal-based development environments.
• Ability to independently develop, debug, and validate scientific computational workflows.
• Strong understanding of physical units, numerical precision, scientific constraints, and reproducibility.
• Excellent analytical, problem-solving, and technical communication skills.

Preferred Qualifications

• Experience with scientific computing libraries such as NumPy, SciPy, pandas, Matplotlib, SymPy, or JAX.
• Familiarity with molecular dynamics, quantum chemistry, finite-difference methods, Monte Carlo simulations, optimization, or statistical modeling.
• Experience with domain-specific tools such as OpenMM, ASE, RDKit, Psi4, LAMMPS, or GROMACS.
• Familiarity with Docker, Conda, Git, continuous integration systems, and automated testing frameworks.
• Experience with high-performance computing (HPC) systems or computationally intensive scientific workloads.

Additional Advantages

• Experience evaluating AI coding agents or terminal-based AI systems.
• Background in research software engineering or scientific benchmark development.
• Experience creating automated graders and objective evaluation frameworks.
• Publications or open-source contributions in computational science or related research areas.
• Experience converting experimental or research workflows into reproducible computational packages.

Contract and Compensation Details

This is a 5-week, full-time contractor assignment requiring 40 hours of work per week and at least 4 hours of daily overlap with PST.

Compensation follows a pay-per-task model, with $300 paid for each approved task that meets the project's specifications and acceptance criteria. Each task is expected to take approximately 6–8 hours on average, depending on complexity.

The contract does not include medical benefits or paid leave. Candidates must be available to start immediately and commit to the required schedule throughout the engagement.