Robotics RL Engineer - Sim-to-Real & Production

Roboticsconnected; Fremont (CA); On-site; USD 120,000 - 190,000; Full time

Job summary

Dexmate is actively seeking reinforcement learning experts to design and deploy RL algorithms that enhance our robotic platforms. You will work on manipulation, locomotion, and navigation tasks, leveraging simulation and real-world data to push the state of the art.

You will collaborate with robotics engineers, scale training pipelines, and help bring RL models into production as part of a fast-paced, open ecosystem focused on democratizing physical AI.

Qualifications

• Strong experience with reinforcement learning algorithms (PPO, SAC, TD3, DDPG, etc.).
• Hands-on experience with robotics systems and real-world or simulated environments.
• Proficiency in Python and DL frameworks such as PyTorch, TensorFlow, or JAX.
• Solid understanding of robot kinematics, dynamics, and control.
• Experience with GPU-based robotics simulation tools like Isaac Gym, Isaac Lab, or SAPIEN.

Responsibilities

• Design and implement reinforcement learning algorithms for robotics tasks
• Develop and optimize RL training pipelines in simulation and real-world settings
• Collaborate with robotics engineers to integrate RL models into production systems
• Conduct experiments to evaluate and improve algorithm performance
• Scale training infrastructure for learning across multiple robots

Skills

Reinforcement learning

Python

Deep learning frameworks

Robotics systems

Robot kinematics and control

GPU-based simulation

Tools

Isaac Gym

Isaac Lab

SAPIEN

Job description

Dexmate is actively seeking reinforcement learning experts to design and deploy RL algorithms that enhance our robotic platforms. You will work on manipulation, locomotion, and navigation tasks, leveraging simulation and real-world data to push the state of the art.

You will collaborate with robotics engineers, scale training pipelines, and help bring RL models into production as part of a fast-paced, open ecosystem focused on democratizing physical AI.