Own the perception-to-action pipeline: fine-tune VLA models, collect imitation-learning data, transfer from simulation to the real robot, and add touch to the policy.
- Design and train Vision-Language-Action / VLSA models for dexterous manipulation and pick-and-place tasks
- Build and maintain imitation learning pipelines from teleoperation data
- Develop and optimize simulation environments in Isaac Sim, Isaac Gym, or MuJoCo
- Lead sim-to-real transfer efforts (domain randomization, system identification, policy adaptation)
- Integrate stereo depth perception (ZED 2i or equivalent) into action models
- Implement and refine SLAM-based navigation for structured warehouse environments
- Establish evaluation frameworks for accuracy, generalization, and cycle time performance
- Deploy trained policies onto physical robotic systems and iterate based on real-world failure modes
#J-18808-Ljbffr
This is a strong role to take on the Founding AI / Robotics Engineer role at Cerebel.
This posting is for the Founding AI / Robotics Engineer role at Cerebel, based in Plano, TX, United States.