About the Role

As a Simulation Engineer, you will own the simulation-to-reality loop that turns our 3D assets into training-ready robotics environments. You'll create high-fidelity environments in simulation, train and evaluate policies in simulation, and critically validate that what works in sim transfers to the real world. Working across simulators like NVIDIA Isaac Sim and physical robot setups, you'll close the real2sim2real gap that determines whether our assets and environments are genuinely useful to robotics customers.

Responsibilities

• Build, configure, and optimize simulation environments in robotics simulators (e.g., NVIDIA Isaac Sim, Isaac Lab, or equivalent).
• Train and evaluate robot policies in simulation (RL, imitation learning, or task-specific controllers), and design evaluation protocols that measure sim performance meaningfully.
• Validate sim results against real-world hardware — designing and running real2sim2real workflows to measure and close the sim-to-real gap.
• Build real2sim pipelines that reconstruct real environments and assets in simulation, and quantify fidelity against ground-truth captures.
• Ensure generated 3D assets simulate correctly — validating articulation, joint behavior, collision handling, physics stability, materials, and lighting.
• Diagnose and resolve simulation issues such as mesh self-collision, articulation instability, and cross-platform rendering/lighting discrepancies.
• Collaborate with ML engineers, researchers, and 3D artists to make assets simulation-ready and to feed real-world validation back into asset generation.
• Develop tooling and automation to streamline asset ingestion, articulation, training, and validation at scale.

Requirements

• Strong hands-on experience with physics-based simulation, ideally in a robotics setting.
• Direct experience with robotics simulators such as NVIDIA Isaac Sim / Isaac Lab, MuJoCo, PyBullet, or Gazebo.
• Demonstrated experience training and evaluating policies in simulation (reinforcement learning, imitation learning, or equivalent).
• Practical experience with sim-to-real transfer and real2sim2real validation — including domain randomisation, system identification, and measuring/closing the reality gap.
• Experience validating on real robot hardware, and comfort working across the sim/real boundary.
• Solid understanding of 3D geometry, meshes, rigging/articulation, collision, and materials (USD, URDF, or similar formats).
• Proficiency in Python; familiarity with 3D tooling such as Blender or Houdini is a strong plus.
• Understanding of rigid-body dynamics, kinematics, and common simulation failure modes (instability, collision artefacts, articulation errors).