Robotics
Bots n' Beans
We are looking for a Robotics Simulation Software Engineer to build and improve Field AI’s internal simulation tools used for autonomy development, debugging, and evaluation.
This role is for someone who goes beyond simply running simulators. We are looking for an engineer who can maintain and extend simulation systems, improve code quality and fidelity, debug simulator-to-robot gaps, and help turn real-world robot behavior into reproducible simulation scenarios.
You will work closely with autonomy, infrastructure, and field teams to make simulation a practical part of the development loop: reproduce field issues, improve autonomy, and validate robot behavior before returning to the field.
Maintain, refactor, and improve internal robotics simulation tools and infrastructure.
Improve simulation fidelity for autonomy evaluation, including robot behavior, sensor modeling, timing, terrain interaction, and environment representation.
Build workflows for scenario execution, log replay, regression testing, metrics, and visualization.
Develop real-to-sim workflows that turn field logs, reconstructed environments, and failure cases into reproducible simulation scenarios.
Work with autonomy and field teams to identify and debug gaps between simulated and real robot behavior.
Improve developer workflows around simulation performance, reproducibility, debugging, and test automation.
Contribute to the long-term simulation strategy for legged robots and humanoids operating in complex real-world environments.
3+ years of experience developing robotics simulation tools, simulation infrastructure, autonomy evaluation systems, or related robotics software.
Strong software engineering skills in C++ and/or Python.
Experience with simulation platforms such as Isaac Sim, Gazebo, MuJoCo, Unity, Unreal, or custom robotics simulators.
Familiarity with ROS2 or similar robotics middleware, including sensors, transforms, logs, and timing.
Experience with robot or simulation assets such as URDF, SDF, meshes, dynamics, or sensor models.
Ability to debug simulator-to-robot gaps across sensing, timing, dynamics, and autonomy behavior.
Comfortable working across research code, production robotics software, simulation infrastructure, and real-world robot deployment workflows.
