Simulation and Digital Twin Lead
GRAM, San Francisco Bay Area
About the company
GRAM Corporation scales skilled labor for the physical economy. We design, manufacture and operate insectoids, general-purpose machines not built in our image. They work in large populations on the critical infrastructure nations depend on. The electric grid comes first. Our research is in how those populations coordinate and keep themselves running. The goal is Self-Replication (SR), the point where a population of machines delivers more labor than it takes to build and run.
The role
You will own the simulation stack and tooling used by the whole engineering team, and build a high-fidelity digital twin of the robot that is validated against real hardware. The Controls, Autonomy and Systems Integration teams will rely on it to develop and test before anything runs on the robot. This is a hands-on role: you will choose the tools, build the models yourself and prove that simulation matches the machine. You start as an individual contributor, with a path to build and lead a team as the company grows.
What you will do
- Own the simulation stack and tooling used by the whole engineering team, including the choice of physics engine.
- Build the digital twin of the robot, from actuators and joints to full body contact, and validate it against hardware data.
- Own sim to real: system identification and comparison of field logs against simulation.
- Build environment simulation of the sites the robot will work in, such as towers and substations, from survey data.
- Simulate sensors such as cameras, depth and IMU to support perception and motion planning.
- Make the simulation run at scale, with many robots in parallel, for policy training and regression testing.
What you bring
- Deep hands on experience with physics simulation of articulated robots, using a robot simulator such as Isaac Sim or Gazebo, or a physics engine such as MuJoCo or PhysX, including how the contact model works.
- You have closed the sim to real loop on real hardware, making simulation match a physical machine through system identification and validation against logged data.
- Experience modeling actuators, transmissions and joints in simulation.
- Strong Python and working C++.
- 5+ years of experience. Depth in physics simulation and sim-to-real on real hardware matters more than years.
Nice to have
- Experience with legged robots or robots on outdoor terrain.
- 3D reconstruction of real sites from photogrammetry or similar survey data.
- Large scale parallel simulation for policy training, such as Isaac Lab or MJX.
- Autonomous vehicle simulation backgrounds are welcome if you can demonstrate work with actuators and joints.
Why join
- A clean slate: define the simulation stack the whole engineering team builds on.
- Early equity, included in the package.
- The chance to own a domain and build a team around it as the company scales.
- A hands on culture where engineers work directly on the core technology.
