Reinforcement Learning Engineer, Grasping

About this role

Persona AI seeks a Reinforcement Learning Engineer to develop dexterous grasping policies for humanoid robots. You'll train RL agents in simulation and deploy them on real hardware, focusing on sim-to-real transfer and complex manipulation tasks.

What you'll do

• Train and refine RL policies for grasping, tool use, and in-hand manipulation tasks
• Develop sim-to-real transfer pipelines using MuJoCo and Isaac Lab simulators
• Design reward functions and curriculum strategies for complex grasping behaviors
• Test and debug policies on physical robotic hands in real-world conditions
• Integrate tactile sensing and feedback into grasp control systems
• Evaluate and benchmark grasping performance across diverse objects and scenarios

What they're looking for

• Reinforcement learning (RL) for robotic manipulation
• Python and deep learning frameworks (PyTorch, JAX)
• RL libraries (rsl_rl, skrl) and simulation environments
• Sim-to-real transfer techniques and domain randomization
• MuJoCo and Isaac Sim experience with mesh/collision preparation
• Hardware testing and real-world policy validation
• Reward shaping and policy evaluation methodologies
• Contact-rich manipulation and tactile sensor integration

Benefits

• Competitive compensation with performance-based bonus
• 99% employer-covered medical benefits
• Early-stage equity
• Competitive paid time off
• Paid company winter break (December 24 - January 2)
• Access to advanced hardware labs and cutting-edge robotics tools