Teams at Flexion Robotics

Develop robust, real-time state estimation and sensor fusion for humanoid robots using proprioceptive and visual sensors. Implement optimization- and filtering-based estimators and learned odometry, combine classic methods with modern ML (seq2seq, autoregressive, diffusion), deploy C++ systems, and train models in Python/PyTorch while working hands-on with robotic or AR/VR hardware.

The Research Engineer will develop VLA models for humanoid robots, focusing on dexterous manipulation and bridging human actions with motor control using large-scale video datasets.

The Software Engineer will design, build, and maintain the simulation infrastructure for robot learning, focusing on simulation fidelity, training efficiency, and software validation while collaborating with cross-functional teams.