Simulation Engineer, Autonomous Vehicle Operations
Build the simulation environment where Autzu tests dispatch, positioning, charging and service logic against real operating data before autonomous vehicles run it on the street.
Autzu builds autonomous infrastructure for autonomous operators: the software that deploys, monitors and orchestrates autonomous vehicles across vehicle platforms, ride platforms and markets. We hold years of real trip, reservation and telemetry data. You'll turn that data into replayable scenarios and digital twins of the cities and sites our vehicles work in, using NVIDIA Omniverse, Isaac Sim and Cosmos, so every change to orchestration and exception handling is scored in simulation first. You'll work directly with the Head of Engineering and the optimization and vehicle integration engineers. Not in scope: validating the driving stack itself. We simulate operations around the vehicle, not the driver.
What you'll do
- Build the replay and scenario pipeline that turns historical trips, telemetry and incidents into repeatable simulation runs.
- Model sites, service bays and city operating zones as digital twins in NVIDIA Omniverse and Isaac Sim (OpenUSD).
- Use NVIDIA Cosmos world models to generate rare and edge-case scenarios that our own data does not cover.
- Simulate vehicle state, charging, cleaning and service handoffs so orchestration logic can be tested end to end.
- Define the pass/fail metrics for each release and run simulation as a gate in CI.
- Keep simulation honest: compare simulated results with what happened in live operations and close the gaps.
What we're looking for
- 4+ years building simulation for autonomous vehicles, robotics or large logistics systems.
- Hands-on with at least one of NVIDIA Isaac Sim, Omniverse, Cosmos, CARLA, or an equivalent AV or robotics simulator.
- Strong Python and C++; comfortable with OpenUSD, ROS 2 and GPU compute.
- Experience building scenario generation, log replay or hardware-in-the-loop test pipelines that a team relied on.
- Based in the San Francisco Bay Area (hybrid).
- Nice to have: synthetic data generation, discrete-event or traffic simulation (SUMO), reinforcement learning in Isaac Lab, NVIDIA cuOpt.
