Robotics Simulation Engineer

About the role

AIRoA collects large-scale real-world robot data, including from humanoid robots, and develops generative AI foundation models for robotics. It is a project operator for a data platform funded by METI and NEDO, aiming to build an open Robot Data Ecosystem. This role designs and develops simulation environments, evaluation scenarios, benchmarks, log replay and automated testing infrastructure for humanoid and service robots, and contributes to improving models, sensors and control systems through synthetic data and sim-to-real analysis.

Key responsibilities

• Design and implement simulation environments, scenarios, evaluation tools and log replay functions for humanoids, mobile and service robots.
• Set up simulation-based software testing, automated regression testing and CI/CD integration to catch performance and safety regressions early.
• Build a scenario library and benchmark suite covering real-world tasks, failures, sensor anomalies, contact, crowded environments and state transitions.
• Prepare synthetic data, sensor simulation, domain randomization and log-derived scenarios for perception, navigation and VLA evaluation.
• Analyze the difference between physical robot logs and simulation results and improve models, sensors, control systems, environment representations and metrics.

Required profile

• Professional experience with robotic arms and robotic hands.
• Experience developing and running robotics, factory automation or physics simulation environments.
• Experience designing systems used by development teams such as test cases, benchmarks, CI pipelines and automated reporting.
• Ability to turn knowledge of sensors, control systems, navigation, manipulation or hardware constraints into simulation requirements.

Required skills

• Hands-on experience with simulators such as NVIDIA Isaac Sim and MuJoCo.
• Python for evaluation pipelines, data processing, simulation tools, log analysis, visualization and test automation.

Preferred: organizing requirements across autonomy, VLA, navigation and hardware teams; cloud-based simulation, GPU acceleration, containers and distributed execution; imitation learning, reinforcement learning, VLA evaluation or ROS/ROS2, Nav2, MoveIt experience; Gaussian splatting, NeRF or neural rendering.