Reinforcement Learning (RL) Engineer

About the role

We are seeking a Reinforcement Learning (RL) Engineer to design, train, and ship learning‑based control policies for our robotic platforms. You will own the end‑to‑end training loop, from simulation and logged data to real‑world evaluation, collaborating closely with simulation, perception, and robotics teams.

Key responsibilities

• Design, implement, and maintain RL training pipelines for manipulation, navigation, and whole‑body control.
• Develop and tune policies in simulation and on hardware, establishing clear benchmarks for success, robustness, and regression detection.
• Integrate RL stacks with Vision‑Language‑Action (VLA) systems, planners, low‑level controllers, and deployment interfaces.
• Create reward functions, curriculum learning, and domain randomization strategies to improve sim‑to‑real transfer.
• Own dataset and experience pipelines for offline RL, imitation, and hybrid training.
• Build evaluation harnesses in simulation and on physical robots, analyze failures, and drive iterative improvements.
• Collaborate with simulation engineers on environments and synthetic data, and work with software/embedded teams on inference deployment and safe rollout.

Required profile

• Degree in Robotics, Computer Science, Electrical Engineering, Machine Learning, or equivalent industry experience.
• Proven track record in reinforcement learning for control or embodied AI (published work or shipped systems).
• Strong ability to debug end‑to‑end pipelines from training curves to real‑robot execution.
• High agency, disciplined experimentation, and comfort working across ML and robotics domains.

Required skills

• Python programming.
• Deep learning frameworks: PyTorch (preferred), JAX.
• Physics simulators: Isaac Lab / Isaac Sim, MuJoCo, Brax, Gazebo‑based stacks.
• Reinforcement learning concepts: MDPs, policy gradients, actor‑critic methods.
• Robot kinematics, dynamics, and control interfaces (position, velocity, torque).
• ROS / ROS 2, MoveIt, motion‑planning integration.