Key Skills

Job Description

Author research-grade AI evaluation tasks from your own computational work. Open to PhD students, postdocs, research scientists, and research-active Master's students. AfterQuery is building a network of researchers to author and validate research-grade scientific-computing tasks for AI models. A strong task is a real piece of computational work from your own research, something you've actually run, debugged, or been paid to do: simulation setup, numerical solvers, model fitting, data reduction, analysis pipelines, or inverse problems. Each task has a main problem, a few connected subproblems, a reference solution, and tests that check the scientific result. The difficulty should come from the science itself, not from withheld information. We're especially looking for researchers in: bioinformatics and computational biology; quantum physics and computational astrophysics; materials science and molecular/quantum chemistry (including electronic structure, polymers and soft matter, chemical kinetics and spectroscopy); condensed-matter and statistical physics; particle and plasma physics; and geochemistry, remote sensing, oceanography, and coastal modeling. Requirements Education: Currently pursuing or holding a PhD with active hands-on research, or a Master's (current or completed) with a research-active record: thesis research, an RA or lab role, a publication, or a clear PhD trajectory. Technical: Working proficiency in Python, enough to run and debug your own computational work. Contract and pay Flexible, independent-contractor work: $200 per approved task, about 2 hours each, with no cap on how many tasks you take on. Tasks go through review, and you may also review other experts' tasks against a short rubric. All accepted experts undergo standard background and identity verification. Responsibilities Design and author original research-grade evaluation tasks based on your own computational work Specify the metrics a strong solution should be judged by, without publishing the exact passing thresholds Revise tasks based on structured review feedback until they're approved Optionally review and calibrate tasks submitted by other experts in your domain Required Qualifications Currently pursuing or holding a PhD with active hands-on research, or a Master's with a research-active record: thesis-based enrollment, an RA or lab role, a real publication, or a clear PhD trajectory Working proficiency in Python, enough to run and debug your own computational work Able to turn your own computational research work into a well-specified evaluation task Able to work independently and manage your own time. This is flexible, task-based work with no fixed weekly schedule Preferred Qualifications Current PhD student, postdoc, or faculty actively publishing in your field Demonstrated experience in scientific computing, data analysis, simulation, or numerical methods Prior experience in data annotation, data labeling, or AI/ML evaluation work Why apply Work on your own schedule: task-based, not a fixed hourly commitment Contribute directly to how AI models are evaluated on real, expert-level scientific work Paid, flexible work: $200 per approved task (about 2 hours each), no cap on how many you take on

Core Responsibilities

Design original, research-grade AI evaluation tasks based on computational work you have personally performed, including defining evaluation metrics and providing reference solutions and tests. Revise tasks in response to review feedback and optionally review or calibrate tasks submitted by other experts.

Requirements

Applicants must be pursuing or hold a PhD with active hands-on research, or have a Master's degree and a research-active record such as thesis work, a research assistant role, a publication, or a clear PhD trajectory. They must be proficient enough in Python to run and debug their own computational work, able to specify evaluation tasks, and able to work independently.