Robotics Reinforcement Learning Engineer (Reinforcement Learning Sim2Real)

About Asiacruit

At Asiacruit we connect top talent with forward-thinking organizations across industries. Our mission is to help businesses grow through smart strategic and people-focused solutions. We support companies by providing high-quality talent for local and global markets.

Position Overview

We are looking for a Robotics Reinforcement Learning Engineer (Reinforcement Learning / Sim2Real Engineers) to develop train and deploy reinforcement learning policies that enable legged robots to move reliably across complex and unpredictable environments.

You will work at the intersection of robot learning locomotion control physics-based simulation and real-world deployment. Your primary focus will be training robust locomotion policies in simulation and transferring them to physical quadruped robots helping them maintain balance adapt to changing terrain and respond safely to dynamic conditions.

This role combines hands-on reinforcement learning development with simulation engineering and low-latency deployment on robotic hardware.

Key Responsibilities

• Develop and train reinforcement learning policies for quadruped locomotion using NVIDIA Isaac Lab Isaac Gym RaiSim or comparable robotics simulation environments.

• Train large-scale parallel locomotion policies across diverse terrain conditions including uneven surfaces varying friction and dynamic disturbances.

• Implement blind and vision-assisted locomotion policies to enable robots to navigate environments with different levels of perceptual input.

• Design and refine domain randomization strategies covering surface friction motor latency payload variations mass distribution and other physical parameters.

• Improve Sim-to-Real transfer to help policies trained in simulation perform reliably on physical robotic hardware.

• Develop and optimize reinforcement learning algorithms including Proximal Policy Optimization (PPO) using PyTorch.

• Deploy trained neural network policies to robotic systems using C ONNX Runtime LibTorch or other suitable inference frameworks.

• Optimize policy inference latency and execution frequency to support responsive real-time locomotion control.

• Integrate foot-contact estimation slip detection and reflex mechanisms to improve stability and protect robotic joints during unexpected contact conditions.

• Evaluate locomotion performance through simulation benchmarks and real-world testing identifying failure modes and implementing iterative improvements.

• Collaborate with robotics controls simulation and embedded systems engineers to integrate learned policies into end-to-end robotic autonomy systems.

Qualifications

• Bachelors or Masters degree in Robotics Computer Science Electrical Engineering Mechanical Engineering Control Systems or a related field.

• Approximately 35 years of relevant experience in reinforcement learning robot learning locomotion control or autonomous robotics.

• Strong programming skills in Python and practical experience with PyTorch.

• Experience implementing reinforcement learning algorithms particularly PPO or other policy optimization methods.

• Hands-on experience with NVIDIA Isaac Lab Isaac Gym RaiSim MuJoCo or comparable physics-based robotics simulators.

• Understanding of legged robotics locomotion dynamics robot balance or contact-rich control.

• Practical knowledge of domain randomization system identification or other techniques for improving Sim-to-Real transfer.

• Experience with C and deploying machine learning models on robotic hardware or edge computing platforms.

• Familiarity with ONNX Runtime LibTorch or other low-latency model inference frameworks is an advantage.

• Understanding of foot-contact estimation slip detection motor dynamics or real-time control is a plus.

• Strong analytical and troubleshooting skills with the ability to investigate simulation failures and validate solutions on physical robots.

Why Join

• Work on advanced robotics technologies at the intersection of reinforcement learning locomotion and robot autonomy.

• Develop learning-based control policies for quadruped robots operating in challenging real-world environments.

• Bridge the gap between simulation-based training and physical robotic deployment.

• Collaborate with multidisciplinary engineering teams on complex robotics research and development.

• Contribute to the development of more adaptive robust and autonomous robotic systems.

Work Location & Arrangement

Hybrid in South Korea. Candidates should be comfortable collaborating with robotics engineering teams and participating in on-site activities as required by the role.