Real-Time Robotics Perception Engineer (C++, ROS 2, Computer Vision)

Your code closes the loop on a robot that walks to the steel and welds it.

About the company

Our client is a venture backed Physical AI company building autonomous robotic welding. Their systems are already running in production at real factories, not pilots, driven by their own AI model for welding. Now they are taking it mobile: a robot arm on a legged platform that walks to the workpiece, scans the seam, and welds on site, built for shipyards and large scale fabrication. Launch customers are lined up and the first units ship in 2027. The team is a couple hundred people and growing fast in Columbus, Ohio.

The role

We are hiring two Real-Time Robotics Perception Engineers for the mobile welding robot team. You will own the vision a real time control loop closes on. The camera sees the seam, your pipeline turns it into geometry, and the robot moves the torch in response, on embedded hardware, at rate, inside a latency budget you set and measure.

What you will do

• Own the perception pipeline end to end: sensing, extraction and geometry, through to the output the control loop acts on
• Get reliable measurements out of imagery the welding arc is actively ruining: glare, spatter, smoke and reflective metal
• Build estimates that know when they are wrong and say so early enough to matter, because the loop will act on a confident wrong answer
• Register partial, imperfect data against a known model, and know when the geometry does not support the answer
• Choose the sensors: run the experiments, compare the options and defend the decision with numbers
• Set technical direction as the perception work grows toward learned methods

What you bring

• You have closed a real time loop on what a camera saw, on real hardware, and something physical moved in response. You can state its rate, its measured end to end latency and how you measured it
• First principles depth in classical vision and 3D geometry: camera models, calibration, triangulation, registration and where it fails, and proper error propagation
• Real time segmentation or feature extraction on degraded imagery, deployed rather than benchmarked
• Strong, modern, production C++ and ROS 2
• GPU accelerated inference on embedded hardware under a hard per frame budget, measuring the tail, not the mean
• An accuracy claim you can defend, measured against an independent reference, with bias reported separately from repeatability

Nice to have

• Weld seam tracking, arc imaging, laser line or structured light sensing
• Vision under harsh conditions: surgical smoke and specular tissue, laser processing, outdoor or field robotics
• NVIDIA Jetson, CUDA and TensorRT in production
• Learned segmentation of thin, low contrast structures

Why this role

• Your work runs on real robots doing real welding, with customers waiting for it
• Ownership of a core system end to end, not a slice of someone else’s architecture
• Pay set at NY/SF rates with Columbus cost of living
• Up to $15,000 relocation reimbursement through the State of Ohio
• H1B transfers and green card support
• Daily free lunch, flexible PTO, comprehensive medical, dental and vision, 6 weeks fully paid parental leave (up to 14 weeks for birthing parents), 401(k)

Location

On site [Various Locations in the US Available]. Relocation support available.