Mechanical Design Engineer - Robot Parts & Enclosures (Contract)

OS3 Robotics is a San Francisco-based robotics startup building general-purpose robotic systems for real-world commercial environments. We are a small, hands-on team working across AI model training, reinforcement learning, mechanical design, electronics, controls, and perception with robots designed, manufactured, and tested in-house.

We design and build our robots in-house, from structural parts to exterior covers, and iterate on them continuously as the robots run in the real world. You'll own the CAD design of robot parts and enclosures — working remotely, with your designs printed, machined, and tested on real robots in our San Francisco lab.

You will:

• Design robot parts, enclosures, and covers in CAD for 3D printing, CNC, and sheet metal
• Package cameras, sensors, compute, actuators, and wiring into clean, serviceable assemblies
• Own designs from concept through prototype, assembly feedback, and revision
• Produce CAD geometry clean enough to feed our simulation and testing pipeline
• Iterate fast: you design remotely, we build in SF, and you get feedback from real robots within days

You have:

• Strong CAD fundamentals — parametric modeling, assemblies, tolerancing, DFM
• Hands-on prototyping experience: you've printed, machined, assembled, and fixed your own designs
• A feel for designing around electronics, cabling, and moving joints
• Clear async communication — you can drive a design loop remotely
• Nice to have:
• Experience with robot arms, grippers, or mobile platforms
• Industrial design sensibility for exterior covers and finishes
• Option to convert to full-time. In-person interns are welcome to apply as well, provided they have CAD experience.

We build the full stack of general-purpose robotics: custom robot hardware and the AI that runs it. Owning both sides gives us a complete data flywheel — our robots collect data in the real world, that data trains our models, and improved models go back on the robots within days.

Our engineering work spans:

• Training VLAs and world models
• Simulation-based training to reduce the need for real robot data
• Reinforcement learning from real-world robot rollouts, improving policies beyond what imitation alone can reach
• Physics simulation with sim-to-real evaluation gates, and GPU inference infrastructure serving policies to robots in real time
• Real-time control software, teleoperation-based data collection, and safety systems
• Custom mechanical design for CNC, sheet metal, and 3D printing, along with custom electronics