Description
Rohlik is hiring a robotics machine-learning engineer to build a human-video-to-robot-policy training corpus and train the first manipulation policies. The role owns the capture specification, ground-truth hand-pose measurement, evaluation harness, training runs, and technical judgment on whether footage is trainable. It requires 3+ years of full-time robotics experience, PyTorch, computer vision, 3D geometry, camera calibration, pose estimation, SLAM, robot-learning policies, and familiarity with ROS 2. The position is part of a small team focused on autonomous grocery-fulfilment robotics.
