01Home

I’m Arthur Petron. I am a systems designer. I build robotics and reality-aligned, distributed, systems. In my formal work, I follow the constructive principles of intuitionistic logic.

Right now I contract with a few robotics companies around the Bay Area through my own shop, Darthur LLC — hardware from idea to prototype, multi-physics simulation, system design, robotics, ML, world models, and hardware operations (including machine shop management). Reach out if you are interested in working together.

Recently, I worked in Germany leading the robotics team tasked with automating operations at a 20,000 sqft production facility running primarily DMG Mori machining centers. Before that I was at OpenAI, leading mechanical engineering and hardware on the robotic dexterity and modeling work using Shadow robot hands, motion capture, vision, to solve a Rubik’s Cube. That followed a short stint at a digitization company focused on paper records in the SF area (East Bay).

I completed my PhD at the MIT Media Lab (my third MIT degree) in Hugh Herr’s Biomechatronics group, where I built (among several other projects, TAing, and teaching) this thing called the FitSocket, a tissue-stiffness measurement device for fitting prosthetic sockets to amputees in a quantitative way. The throughline has always been the same: trying to get hardware to function well in the unpredictability of the physical world.

At the beginning of the 2020s I realized that I needed better tools for abstraction. This led me toward learning math -- specifically category theory, topos theory, and as much of the foundational material required for this. By no means am I an expert. The more I learn in this area, the more I learn just how little I understand. Regardless, I am fascinated with the results I get when I apply the abstractions from increased mathematical understanding to help make complex systems easier to reason about and more robust. I’m interested in “math that compiles” (read: Lean, Coq, Agda, etc.) into architecture, models, and composed systems that are consistent with what we may measure in physical reality.

Please, consider this website to be an extension of my personal notebook: electromechanical designs, actuator ideas, simulation and control experiments, and math notes and musings as I try to turn abstract structure into working code, written monograms, and hardware. If you care about robotics, simulation, and mathematics of building real systems you’ll probably find something useful here. Some things will be more serious than others.

• Neuroplastic — scientific advising2025
• Cinequest — 12 photographs2026·03
• Quinzaine — 9 photographs2025·05
• Sundance — 41 photographs2025·01
• The 2024 Nobels — 14 photographs2024·12

• GitHubcode
• Google Scholaracademic presence
• MIT Media Labprosthetics work
• arthur-tex-styleinspo for this site design