18 days ago
Buenos Aires, Argentina +49 moreMid Level / Senior
Responsibilities
- Design and build MuJoCo simulation environments for robotics research.
- Implement and tune reinforcement learning algorithms like PPO, SAC, and TD3.
- Define reward functions, observation spaces, and action spaces for training agents.
- Debug and optimize physics simulations, including contact models and actuator dynamics.
- Evaluate trained policies for stability and sim-to-real transfer potential.
- Document environment specifications and training procedures clearly.
- Collaborate asynchronously with research teams and stay updated on advances in robot learning.
Requirements
- Strong hands-on experience with MuJoCo or similar simulation tools.
- Solid understanding of reinforcement learning theory and training pipelines.
- Proficient in Python and ML frameworks such as PyTorch or JAX.
- Experience defining reward functions for complex robotic tasks.
- Familiarity with robot kinematics, dynamics, and control fundamentals.
- Ability to read and write MJCF/XML model files and understand their physics implications.
- Self-directed and detail-oriented, comfortable working independently in an async environment.
- Strong written communication skills for explaining reasoning clearly.
Benefits
- Fully remote work from anywhere on the accepted locations list.
- Flexible hours ranging from 15 to 40+ hours per week based on project needs.
- Weekly payment via PayPal or Stripe.
- Competitive compensation ranging from $30 to $70 per hour based on location and seniority.