Career

Key Responsibilities:

• Design, train, and deploy reinforcement learning (RL) algorithms for various robotic motion tasks

• Develop and optimize simulation infrastructure to enable large-scale policy training for general-purpose robots.

• Collaborate with the other teams to integrate learned RL policies into the robot’s existing control architecture.

• Define evaluation metrics, conduct performance testing, and assess policy effectiveness.

Requirements:

• Strong proficiency in developing production-grade code using frameworks such as PyTorch, TensorFlow, or JAX.

• Solid understanding of both online and offline RL algorithms, including PPO and SAC.

• Experience with simulation platforms such as IsaacGym, MuJoCo, Gazebo, IsaacSim, NVIDIA Omniverse IsaacLab, or Bullet.

• Skilled in hyperparameter tuning, cost function design, and overall, RL training optimization.

• Familiarity with advanced RL techniques like domain randomization, curriculum learning, and reward shaping.

• Proficient in using machine learning evaluation tools such as TensorBoard or Weights & Biases.

Preferred Qualifications:

• Experience in deploying learned policies from simulation to real robotic hardware systems.

• Practical experience with large-scale parallel training frameworks such as IsaacGym.

• Background in training locomotion policies for bipedal or quadrupedal robots.

• Demonstrated success in applying RL to real-world problems or in industrial settings.

Benefits:

Competitive salary and benefits package (Open to salary negotiations).  

Opportunities for professional development and career growth.  

Flexible work arrangements.  

A collaborative and innovative work environment.