Robotics Simulation SME

Insight Global; San Francisco (CA); On-site; USD 120,000 - 170,000; Full time

Job summary

Insight Global is sourcing a Robotics and Physics Subject Matter Expert to work with forward-deployed engineers on MuJoCo environments for meaningful robotics and control problems, including robust models, task interfaces, rewards, graders, and scalable execution.

The role emphasizes hands-on MuJoCo ownership, strong Python skills, and expertise in dynamics, control, environment architecture, and reproducibility to ensure robust evaluation and scalable experimentation.

Qualifications

• MuJoCo ownership with hands-on creation and debugging of custom MuJoCo/MJCF environments.
• Strong knowledge of rigid-body dynamics, kinematics, feedback control, robotics, and contact mechanics.
• Experience designing observation/action spaces, resets, termination, rewards, and evaluation protocols.
• Strong Python programming, testing, and profiling; diagnose simulation, controller, policy, and infra failures.

Responsibilities

• Build MuJoCo environments for robotics and control problems.
• Collaborate with forward-deployed engineers to create robust models, task interfaces, rewards, graders, and scalable execution.

Skills

MuJoCo ownership

Dynamics and control

Environment architecture

Python

Job description

Our client is seeking a Robotics and Physics Subject Matter Expert to work alongside forward-deployed engineers to build MuJoCo environments for meaningful robotics and control problems, including robust models, task interfaces, rewards, graders, and scalable execution.

QUALIFICATIONS

• MuJoCo ownership: Hands-on creation and debugging of custom MuJoCo and MJCF environments—not only packaged tasks.
• Dynamics and control: Rigid-body dynamics, kinematics, feedback control, robotics, and contact mechanics.
• Environment architecture: Observation/action design, reset distributions, termination, reward design, and evaluation protocols.
• Implementation: Strong Python, testing, and profiling; diagnoses simulation, controller, policy, and infrastructure failures.
• Reproducibility: Controls seeds, assets, versions, solver parameters, time steps, and regression trajectories.
• Evaluation: Uses baseline controllers or policies, perturbation tests, and adversarial checks for difficulty and robustness.
• Python: Strong Python programming skills and experience with diagnostics/testing/profiling.

PREFERRED

• RL experience, agent coding, benchmarking and evaluation skills.
• dm_control/Composer, MJX, Gymnasium, manipulation, locomotion, biomechanics, or multi-agent simulation.
• Sim-to-real, system identification, domain randomization, trajectory optimization, or model-predictive control.
• C/C++, JAX, accelerator-backed simulation, distributed rollouts, or large-scale experiment infrastructure.