Part-time Research Collaboration

About Steadyworks

Steadyworks is a data research lab powering the next generation of AI. We partner with frontier AI labs to build expert-grade datasets, reinforcement learning environments, and evaluation infrastructure that improve state-of-the-art models. Our team previously built frontier AI systems at Meta, Google Research, Autodesk, and other leading organizations.

What this looks like

A part-time fellowship for experts who already do the work we're trying to measure — competitive programmers, ML researchers, CAD practitioners, and other frontier-domain specialists. You'll help design tasks, author environments, and grade agent behavior.

• Authoring realistic tasks in your domain of expertise
• Grading agent trajectories and writing rubrics that scale
• Advising on what a good evaluation looks like in your field
• Optional: co-authoring writeups or papers where the collaboration warrants it

Good fits include

• Competitive programmers, ICPC / IOI alumni, or top-rated Codeforces / AtCoder contestants
• Working ML researchers who want to shape the evals for the field
• Practicing engineers in CAD, mechanical, or other technical domains
• PhD students looking for a paid, high-signal side collaboration

Logistics

• Fully remote or hybrid in San Francisco
• Flexible hours; usually 5–15 hours per week
• Paid hourly or per-task depending on the engagement

How to apply

Email careers@steadyworks.ai with anything that would help us understand your work.