Undercontrolroboticsinc Simulation Engineer at Noble Machines building digital twins and simulation assets for robots and industrial environments. Responsible for measuring sim-to-real gaps and optimizing simulation performance.
Responsibilities
Build simulation-ready robot parts, tools, and scenes from CAD, descriptions, and scanned or reconstructed scenes. Create reusable assets with consistent scale, coordinate frames, and structure.
Develop collision representations that preserve task-critical geometry and contact behavior. Choose and tune mesh simplification, convex decomposition, and primitive approximations to balance accuracy, stability, and computation.
Own measurement of the physical sim-to-real gap. Design matched hardware and simulation tests, quantify differences in motion and contact behavior, and use the results to calibrate mass, inertia, friction, and collision models. Work with hardware and robotics engineers to collect reference measurements and track improvements.
Own measurement of the perception sim-to-real gap. Compare rendered and real camera observations under matched viewpoints and conditions, and work with learning and evaluation engineers to quantify the effect on downstream perception or policy performance. Use the findings to guide visual meshes, materials, textures, lighting, and camera configuration.
Profile physics and rendering separately. Optimize geometry and texture budgets, asset reuse, instancing, memory use, scene loading, and simulation throughput while maintaining numerical stability and preserving details that affect robot interaction or visual learning.
Develop programmable task environments, scene variations, randomization, and reset logic for training, data generation, and evaluation. Integrate them with the learning and evaluation teams' pipelines.
Automate asset import, conversion, and validation. Review and improve externally supplied assets, and maintain reproducible scenes that teammates can use and extend.
Qualification
Experience building digital twinsHands-on experience with Isaac SimSolid understanding of 3D geometryEspecially relevant experienceContact-rich manipulationIsaac LabCAD conversion
Required
Experience building digital twins, simulation assets, or physically interactive environments for robotics, industrial simulation, or comparable applications.
Practical understanding of collision detection and contact modeling, including how collision geometry, physical parameters, and solver settings affect behavior, stability, and performance.
Experience optimizing 3D assets and scenes for runtime use, including mesh complexity, materials, textures, rendering cost, and memory consumption.
Hands-on experience with Isaac Sim, Omniverse, MuJoCo, or a comparable physics-based simulation platform. Familiarity with USD and robot descriptions such as URDF, or equivalent formats and workflows.
Solid understanding of 3D geometry, coordinate transforms, rigid-body dynamics, and articulated systems, with the ability to diagnose incorrect geometry, inertial properties, joint configuration, and contact behavior.
Working understanding of how simulation supports robot learning and evaluation, and the judgment to choose the level of visual and physical fidelity a task needs.
Especially relevant experience
Contact-rich manipulation, grasping, articulated mechanisms, or robotics simulation.
Isaac Lab, PhysX, scene generation, or synthetic data pipelines with configurable cameras and annotations.
CAD conversion, photogrammetry, scene reconstruction, or asset processing with Blender, Houdini, or similar tools.
Simulation used for imitation learning or reinforcement learning, including randomized environments and parallel rollouts.
Share your resume and up to two representative simulation projects, asset pipelines, or interactive demonstrations. Describe the parts you owned, how you validated physical or visual fidelity, and a performance tradeoff you measured. Examples that show collision geometry and runtime behavior alongside the visual result are especially useful. Apply through our job board or email people@noblemachines.ai.