Machine Learning Engineer - Simulation Framework

Simulation is essential for Zoox to rapidly iterate on our driving software and hardware, and to validate our safety before we drive in the real world. We create virtual worlds to challenge our robots, from real-world data, entirely novel scenarios, or a combination of both. Our simulations need to run at a huge scale to cover everything that might happen, and to help prove our driving to be safe.

As a Machine Learning Engineer on the Simulation Core Team, you will focus on the intersection of machine learning and synthetic environments within our high-speed, GPU-based simulation framework. Our success depends on you driving ML efficiency while solving complex "sim-to-sim" and "sim-to-real" fidelity gaps, ensuring our safety-critical models train on data that perfectly aligns with physical vehicle behavior.

Develop and optimize our GPU-based simulation framework to support complex machine learning training and validation pipelines.

Apply reinforcement learning concepts to solve complex behavioral and path planning challenges in simulation environments.

Identify and resolve "sim-to-sim" and “sim-to-real” fidelity gaps to ensure parity between high-speed ML simulations, high-fidelity 3D environments, and physical vehicle execution.

Build systems that allow autonomy users to self-serve data generation and accelerate their training iterations.

Write robust, production-ready code to integrate advanced ML algorithms directly into our core simulation architecture.

PhD or Master’s in computer science, robotics, machine learning, or a related field.

Deep understanding of reinforcement learning and its application in simulated or robotic environments.

Hands-on experience developing, training, and fine-tuning deep learning models using modern frameworks (e.g., JAX or PyTorch).

Strong proficiency in C++ and Python for building and deploying production machine learning systems.

Experience analyzing and bridging fidelity gaps between synthetic training data and real-world execution.

Experience with GPU programming (CUDA) or high-performance compute clusters.

Automotive or autonomous robotics industry experience.

Strong background in deterministic systems and latency optimization.