ML Engineer, Behavior Planning

Aurora•Mountain View, CA

12d•$139,000 - $223,000•Hybrid

About The Position

Aurora is seeking an ML Engineer to join the Behavior Planning Team. This is a high-impact role where you will sit at the frontier of autonomous vehicle technology. You won't just be applying AI and ML methods; you will be advancing the state-of-the-art for how a self-driving system reasons about the world, interacts with humans on the road, and masters complex, high-dimensional decision-making.

Requirements

BS, MS, or PhD in Robotics, Machine Learning, Computer Science, or a related quantitative field, or equivalent practical experience.
2+ years of experience developing state-of-the-art ML models, either in a research or production setting.
Hands-on experience working on Imitation Learning or Reinforcement Learning applied to physical or simulated agents.
Experience training large models on massive datasets using distributed computing.
Fluency in Python, with a focus on writing high-performance, maintainable code.
Deep experience with PyTorch (preferred) or another modern ML framework, and a mastery of modern ML architectures including Transformers and Diffusion Models.

Nice To Haves

A track record of publications in top-tier ML conferences (NeurIPS, ICML, CoRL, CVPR, AAAI).
Experience deploying complex ML systems in production environments.
Experience in developing generative models or neural simulators for synthetic data generation.
Fluency with C++, with a focus on writing high-performance, safe, and maintainable code.

Responsibilities

Define the architecture of our onboard planning models: Develop and deploy large-scale models trained with Imitation Learning and Reinforcement Learning that enable the Aurora Driver to navigate complex environments with human-like fluidity and superhuman safety.
Revolutionize evaluation: Develop powerful offboard critic models that can evaluate driving behavior at scale, identifying subtle nuances in comfort, progress, and safety that traditional heuristics miss.
Build our next-generation simulation engine: Architect cutting-edge offboard foundation models that power our simulation engine, creating realistic "world models" to test the Aurora Driver against an infinite variety of edge cases.
Bridge research and production: Reach new frontiers of autonomous driving technology by pushing forward the state-of-the-art, but also deploy your models on real production vehicles that drive on public roads and must meet the highest standards of safety.