Simulation Engineer – Embodied AI

About RoboTensor

RoboTensor is a Physical AI research and development company building the infrastructure that turns robot foundation models into specialized physical capabilities.

Simulation is a core part of that infrastructure. We reconstruct physical tasks and operating environments so capabilities can be developed, trained, evaluated, and improved repeatedly before being proven in the real world.

About the Role

We are looking for a Simulation Engineer – Embodied AI to build the environments where physical intelligence is trained and tested.

You will create simulation environments and digital twins that reproduce real tasks, robots, objects, and operating conditions with enough fidelity to generate useful training data and meaningful evaluations.

This is not primarily a graphics role. Simulation exists to make policies better.

You will work closely with embodied AI and research engineers to determine what needs to be simulated, which aspects of reality matter for learning, and how simulated experience can translate into physical performance.

What You’ll Do

• Build simulation environments for training and evaluating embodied AI policies.
• Reconstruct real robots, workspaces, objects, task conditions, and physical interactions.
• Develop digital twins of real operating environments.
• Create task definitions, success conditions, variations, and edge cases.
• Generate large-scale synthetic training data and trajectories through simulation.
• Build procedural environment and scenario generation systems.
• Develop domain randomization and variation strategies to improve policy robustness.
• Model relevant physics, contacts, sensors, cameras, actuators, and control interfaces.
• Build interfaces between simulation environments and policy-training pipelines.
• Develop scalable systems for running large numbers of simulation episodes.
• Measure and reduce gaps between simulated and real-world behavior.
• Profile and optimize simulation throughput.
• Build tools that make it easy for ML engineers to create experiments and modify tasks.
• Collaborate on benchmarks that accurately measure physical capability.

What We’re Looking For

• Strong software engineering skills.
• Experience building simulation, physics, game-engine, or 3D systems.
• Strong understanding of coordinate systems, rigid-body dynamics, geometry, and physical interaction.
• Ability to build clean, programmable environments rather than one-off visual demonstrations.
• Experience working with Python and a systems language such as C++ is valuable.
• Ability to reason about which aspects of physical reality matter for learned behavior.
• Strong debugging skills across simulation, software, and physical systems.
• Comfortable working closely with machine learning engineers and researchers.

Nice to Have

• Experience with MuJoCo, Isaac Sim, SAPIEN, Bullet, or similar simulation platforms.
• Experience with embodied AI or robot-learning environments.
• Experience with manipulation, grasping, locomotion, or physical interaction.
• Experience generating synthetic data for machine learning.
• Experience with domain randomization or sim-to-real transfer.
• Experience with USD, URDF, meshes, CAD assets, or robot descriptions.
• Experience scaling simulation workloads across GPUs or compute clusters.

What Success Looks Like

Your simulations accelerate learning rather than merely reproducing reality visually.

Training data generated in your environments improves policies. Evaluations performed in your environments predict meaningful differences between models. And improvements developed in simulation increasingly survive contact with the real world.