Robotics ML Expert, AI
About The Position
Design, build, and iterate on MuJoCo simulation environments for robotics research and AI training Implement and tune RL algorithms (PPO, SAC, TD3) to train agents on simulated tasks Define reward functions, observation spaces, and action spaces that produce robust, transferable policies Debug and optimize physics simulations — contact models, actuator dynamics, scene configs Evaluate trained policies for stability, generalization, and sim-to-real transfer potential Document environment specs, training procedures, and experimental results clearly Collaborate async with research teams and stay current with advances in robot learning and embodied AI RLHF in one line: Generate code → expert engineers rank, edit, and justify → convert that feedback into reward signals → reinforcement learning tunes the model toward code you'd actually ship.
Requirements
- Strong hands-on experience with MuJoCo (or via dm_control, Gymnasium-Robotics, or similar)
- Solid understanding of RL theory and practical training pipelines
- Proficient in Python + ML frameworks (PyTorch or JAX)
- Experience defining reward functions for complex robotic tasks
- Familiar with robot kinematics, dynamics, and control fundamentals
- Can read and write MJCF/XML model files and understand their physics implications
- Self-directed, detail-oriented, comfortable working independently in an async environment
- Strong written communicator — a big part of this role is explaining your reasoning clearly
Nice To Haves
- Experience with sim-to-real transfer — domain randomization, system identification
- Familiarity with other physics simulators: Isaac Gym, PyBullet, Drake, or Genesis
- Background in multi-agent environments or hierarchical RL
- Published research or open-source contributions in robotics, RL, or embodied AI
- Experience with imitation learning, model-based RL, or world models
- Graduate-level coursework or degree in robotics, ML, CS, or a related field
Responsibilities
- Design, build, and iterate on MuJoCo simulation environments for robotics research and AI training
- Implement and tune RL algorithms (PPO, SAC, TD3) to train agents on simulated tasks
- Define reward functions, observation spaces, and action spaces that produce robust, transferable policies
- Debug and optimize physics simulations — contact models, actuator dynamics, scene configs
- Evaluate trained policies for stability, generalization, and sim-to-real transfer potential
- Document environment specs, training procedures, and experimental results clearly
- Collaborate async with research teams and stay current with advances in robot learning and embodied AI
Stand Out From the Crowd
Upload your resume and get instant feedback on how well it matches this job.
What This Job Offers
Career Level
Mid Level
Number of Employees
11-50 employees
