Domain Specialists (Science, Operations, Hardware, Media) — Task Ideation

About the Project

We build hard, verifiable tasks that AI agents must solve on a computer: analyze the data, run the simulation, fix the model, produce the report. Each task has an environment with files and tools, an instruction, and a check that a program can run to tell whether the result is correct.

AI agents know the textbook. They still fail where experience matters: the instrument quirk nobody documents, the business rule that only applies at month-end, the result that looks right and isn't. You know where those are.

You don't need to be a software engineer.You work in pairs with a software engineer who builds the environment and the tests. Your part is the expertise: real episodes from your work, what a correct result looks like, and where a smart newcomer would go wrong. Much of this is captured through an AI-guided interview.

Who we're looking for

• Life Sciences:bioinformatician or computational biologist (genomics, proteomics, structural biology) who writes analysis code in Python or R. MSc/PhD or 3–5 years of practice. Nextflow or Snakemake a plus.
• Physical Sciences & Chemistry:analytical or computational chemist, physicist or earth scientist who uses code as a tool. MSc/PhD with lab or simulation practice, currently doing research. Owning unpublished data helps.
• Formal Methods & Algorithms:mathematician or computer scientist: Lean/Coq formalizer, ICPC or olympiad competitor, or operations-research modeler. MSc/PhD in mathematics or computer science.
• Operations:business professional who codes a little, with 5+ years in claims and insurance, regulatory-risk quant, trade compliance, logistics, production planning or marketing analytics. A domain degree or certification; you have automated part of your own job (SQL, Python or advanced Excel).
• Hardware & Robotics:mechanical, electrical or robotics engineer: scripted CAD (FreeCAD, build123d), RTL/FPGA (Verilog), or robotics (Drake, ROS). BSc/MSc in a related field.
• Media:music theorist with a graduate degree (harmony, voice leading), or audio/DSP engineer with an MSc/PhD in MIR/DSP.

Responsibilities

• Share real episodes from your work through AI-guided interviews, with concrete detail (files, tools, data formats, numbers)
• Define what a correct result looks like and how it could be checked by a program
• Work with a software engineer to turn your episode into a task, and review the task for correctness in your field
• Point out where a smart but inexperienced person would go wrong
• Replace any confidential or unpublished data with realistic, made-up examples

Project Details

Remote, task-based work in Vetto Arena. Compensation is paid per approved task.

If you know where AI gets your field wrong, this project is for you.

02 — ELIGIBILITY

Who can apply

• Proven, hands-on experience in one of the fields above, in research or industry
• Part of your real work happens on a computer: data files, analysis scripts, simulations, models, spreadsheets or design tools
• Comfortable working fully in English (reading and writing)

Nice to have

• Publications with code, open-source contributions in your field, or competition results
• Experience automating parts of your own work
• Basic familiarity with the command line