We work on the dexterity problem for humanoids

“Proception is seeking an expert in reinforcement learning to help build foundation models for dexterous humanoid manipulation. You will lead the design of scalable RL pipelines, train robust policies on high-DOF robotic hands, and deploy intelligent behaviors that push the limits of real-world robotic control.”

🎯Role Details

•Design and deploy scalable RL pipelines for high-DOF humanoid robots

•Develop foundation models for manipulation tasks like grasping and in-hand reorientation

•Train policies to perform reliably under real-world uncertainty

•Simulate manipulation environments using MuJoCo or Isaac Lab

•Optimize robot hardware configurations via learning-driven exploration

•Apply advanced RL techniques to model complex object-hand dynamics

•Build imitation learning pipelines for learning from human demonstrations

•Leverage multimodal feedback to enable adaptive manipulation skills

⚡Requirements

•MS or PhD in Computer Science, Robotics, RL, or a related field, or equivalent experience

•Deep expertise in reinforcement learning, policy optimization, and control theory for high-dimensional continuous systems

•Strong background in probability, optimization, and linear algebra

•Experience training and deploying RL models on real robots

•Proficiency in Python and C++ on Linux/Unix systems

•Hands-on with robotics simulators like MuJoCo or Isaac Sim

•Experience with imitation learning or inverse RL methods

•Skilled at writing scalable training code and managing large experiments

•Familiarity with tactile sensing or robot morphology optimization

•Self-driven and thrive in a high-velocity, high-impact environment

💰Compensation & Benefits

•Help build the future of dexterous humanoid manipulation

•Work alongside world-class roboticists and engineers

•Access to state-of-the-art robotic platforms and labs

•Competitive salary with equity

•Comprehensive health benefits

Please also apply on our website https://www.proception.ai/careers for better tracking. Thanks!