Hey all,
Background on us: we’re a venture-backed startup based in Austin, Texas. We’re reinventing the way America builds ships, centered around high-mix robotics.
Our current platform is designed to take a 3D model of part of a ship, automatically generate an assembly sequence, and assemble and weld it using industrial robots and a collection of sensors.
The stuff we’re building is big - currently up to 10x40’ - and will go into everything from commercial ships to autonomous vessels.
The challenge we’re solving: most of the parts are composed of similar primitives, but it’s rare that any two are exactly alike.
To solve this, we’ve built our software stack from the ground up in rust - SIMD-accelerated motion planning, elastokinematic calibration, physics-based simulation - our core robotics libraries have very few external dependencies.
The goal is rapid generation speed (currently <30s) on commodity hardware, high precision, and full control at the actuator level.
Example of our simulator:
And the same software running on real robots in Austin:
Our philosophy is: closed-form first for accuracy & repeatability, and then layer reinforcement learning on top where it makes sense.
The job description is available on our website here:
If this is something you find interesting, don’t hesitate to apply even if you don’t meet the exact criteria. Just be obsessed with robots =D
-Dustin