Robotics ML Expert – MuJoCo Environments

About the Role

What if your expertise in robotics and machine learning could directly shape how the next generation of intelligent agents learn to move, manipulate, and interact with the physical world?

Alignerr is seeking a Robotics ML Expert with deep, hands-on MuJoCo experience to design, build, and refine advanced simulation environments. In this role, you will train AI systems to master complex real-world tasks—spanning everything from locomotion and dexterous manipulation to multi-agent coordination.

This is a fully remote, flexible contract role tailored for experienced practitioners who live and breathe physics simulation, reinforcement learning (RL), and robot control. If you love wrangling MJCF files, tuning reward functions, and debugging contact dynamics, this role was built for you.

What You’ll Do

• Develop MuJoCo Environments: Design, build, and iterate on advanced physics-based simulation environments for cutting-edge robotics research and AI model training.

• Tune RL Pipelines: Implement, optimize, and scale reinforcement learning algorithms (including PPO, SAC, and TD3) to train agents in complex simulated tasks.

• Shape MDP Frameworks: Define robust reward functions, observation spaces, and action spaces that yield highly transferable and stable control policies.

• Optimize Physics Simulation: Identify and debug contact models, actuator dynamics, and complex scene configurations to ensure maximum realism.

• Assess Sim-to-Real Viability: Evaluate trained policies for overall structural stability, generalization capabilities, and physical sim-to-real transfer potential.

• Document Technical Rigor: Clearly and thoroughly document environment specifications, edge-case failure modes, and training metrics for asynchronous research teams.

Who You Are

• MuJoCo Fluent: Possess strong, hands-on experience building environments directly in MuJoCo or via wrappers like dm_control and Gymnasium.

• RL Practitioner: Hold a solid theoretical and practical grasp of reinforcement learning theory, optimization loops, and training pipelines.

• Core Developer: Proficient in Python and deeply comfortable working within major machine learning frameworks like PyTorch or JAX.

• MJCF Expert: Highly capable of reading, writing, and debugging native MJCF/XML model files, with a sharp intuition for their physics implications.

• Controls Aware: Familiar with robot kinematics, rigid body dynamics, and foundational robot control principles.

• Asynchronous Professional: Detail-oriented, self-directed, and entirely comfortable executing technical milestones without close supervision.

Nice to Have (Bonus Points)

• Proven experience with sim-to-real transfer techniques (e.g., domain randomization, system identification).

• Familiarity with alternative physics simulators such as Isaac Gym, PyBullet, Drake, or Genesis.

• Background in multi-agent environments, hierarchical RL, imitation learning, or world models.

• Published academic research or meaningful open-source contributions in robotics, RL, or Embodied AI.

Why Join Us?

• Impact Embodied AI: Directly influence how autonomous agents transition from digital simulations to successfully navigating the physical world.

• Total Freelance Autonomy: Enjoy a 100% remote, milestone-driven framework that allows you to work whenever and wherever it suits your schedule.

• Elite Global Network: Collaborate asynchronously alongside tier-one AI research labs and an international community of top-tier ML practitioners.

• Pipeline Scalability: Gain exposure to state-of-the-art architectures with long-term potential for contract extensions as new projects launch.