Helix AI Engineer, Figure

• Hands-on experience developing and deploying robot learning systems on real robots
• Strong background in robot manipulation and visuomotor control
• Experience with behavior cloning, reinforcement learning, or related learning-based manipulation methods
• Proficiency in Python and/or C++ for robotics and ML systems
• Experience with modern deep learning frameworks (e.g., PyTorch)
• Ability to design experiments, analyze failures, and iterate quickly in real-world robotic systems
• Solid understanding of the tradeoffs between classical robotics approaches and learning-based methods
• Thrive in fast-paced, ambiguous environments where solutions require exploration and ownership

Desirable

• We are looking for a Helix AI Engineer, Robot Learning with a strong robotics learning background to help develop and improve our visuomotor manipulation policies, with a heavy emphasis on real-robot deployment
• Design, train, evaluate, and deploy learning-based visuomotor policies for humanoid robot manipulation
• Develop manipulation behaviors such as grasping, pick-and-place, object reorientation, door opening, bimanual manipulation, and basic assembly
• Apply and extend techniques including behavior cloning, reinforcement learning, and VLA reasoning
• Train models that are robust to real-world challenges such as sensor noise, partial observability, contact dynamics, and environment variability
• Own the full pipeline from data collection on real robots to model training, evaluation, and deployment
• Work closely with simulation and digital twin tooling where useful, while prioritizing real-world performance and transfer
• Collaborate with perception, controls, systems, and hardware teams to integrate policies into a full autonomy stack
• Evaluate tradeoffs between learning-based and classical approaches and make principled design decisions
• Write high-quality, well-tested software that ships to and runs reliably on physical humanoid robots
• Partner with integration and testing teams to continuously improve robustness, performance, and deployment velocity