Embodied AI Simulation Engineer

Job Description:

Core Objective: Build the digital pipeline from the real world to simulation space. Reconstruct and calibrate visual and physical properties of robots, objects and scenes, and deliver SimReady interactive assets to reduce the sim-to-real gap, supporting algorithm training, simulation evaluation and policy transfer to real hardware.

Responsibilities

• Own the end-to-end Real2Sim workflow based on Isaac Sim, MuJoCo, Genesis, Blender and other simulation platforms, including 3D reconstruction, appearance reconstruction, dynamic identification, calibration of contact, friction and inertia parameters, to align dynamic and perceptual behaviors between simulation and real robots.

• Create, validate and iterate SimReady assets following USD / URDF / MJCF standards. Configure joints, collision shapes, mass/inertia, joint limits, semantic labels and affordance properties for robots and interactive objects, to achieve physical parameter calibration of simulation assets, ensuring stable usability in simulation training.

• Process multi-modal real-world captured data: camera calibration, point cloud & mesh processing, texture and PBR material calibration. Leverage 3DGS / NeRF and related reconstruction methods to digitize real scenes and export assets compatible with simulation engines.

• Develop automated tools for calibration, data processing and asset quality inspection. Document asset specifications, calibration workflows and evaluation metrics to improve pipeline throughput and sim-real alignment accuracy.

• Collaborate with perception, robot learning, simulation engine and hardware teams, design quantitative sim-real benchmark experiments, identify simulation bias, and continuously improve calibration protocols and asset standards for closed-loop data iteration.

• Design excitation trajectories for robot system identification, collect time-series dynamic data from physical robots, calibrate and reproduce non-ideal dynamic behaviors including nonlinear joint friction and transmission backlash.

• Collaborate with modeling, data, control and hardware teams to leverage the simulation system for evaluation, data generation, policy learning and product capability iteration.

Requirements

Must-have

• Skilled in Python and C++, with the ability to independently implement calibration scripts, engine plugins and data processing utilities.

• Proficiency in at least one simulation/graphics tool such as Isaac Sim, MuJoCo, Genesis or Blender, with solid understanding of computer graphics and rigid-body dynamics. Familiar with physics simulation engines such as PhysX and Newton.

• Hands-on experience with Real2Sim pipelines and SimReady asset standards; capable of independently building, configuring and validating robot or object assets including joint setup and physical property tuning.

• Knowledge of 3D reconstruction, camera imaging, point cloud processing and camera calibration; fundamental understanding of system identification, parameter calibration and contact/friction models for physical simulation. Familiar with PBR rendering and material creation.

• Ability to decompose requirements, implement technical solutions, run comparative experiments and root-cause sim-real mismatches from visual reconstruction or physical simulation.

• Capable of designing dynamic system identification experiments, analyzing measured hardware data, and quantifying sim-real errors introduced by nonlinear dynamics.

• Proficient in data analysis, experimental design and engineering delivery; capable of independently constructing simulation, evaluation, data and feedback pipelines.

Nice-to-have

• Proven track record delivering Real2Sim projects, especially for robotic arms, dexterous hands or humanoid robots.

• Deep familiarity with USD / URDF / MJCF, experience building asset libraries and automated asset validation pipelines.

• Experience with 3DGS, NeRF, PBR reconstruction and photogrammetry.

• Background in system identification, excitation trajectory design, dynamic parameter estimation and contact model tuning.

• Familiar with nonlinear friction models such as LuGre and Stribeck, with experience modeling and tuning joint hysteresis and backlash.

• Experience with motion capture, multi-sensor calibration and real robot hardware experiments.

• Hands-on experience with VLA, World Model, Diffusion Policy, Offline / Online RL and Agentic Learning.

• Solid experience in Sim-to-Real, Real-to-Sim, System Identification, Domain Randomization, Residual Dynamics and Differentiable Simulation.

Default Benefits:

• Health insurance

• Vision care

• Dental coverage

• 401(k)

• Paid holidays

• PTO (Paid Time Off)

• Sick leave