Zoox

The Perception team at Zoox is at the forefront of leveraging GenAI to create synthetic data, unlocking scalable training and evaluation for our autonomous system's perception and entire stack. As a Generative AI Engineer, you will develop and train cutting-edge models for sensor-level scenario generation, utilizing world models and radiance fields techniques with large-scale proprietary data. This role directly impacts the productivity, safety, and capabilities of Zoox's autonomous system by validating algorithms in real-world conditions.

In this role, you will:

• Define and execute the ML roadmap for synthetic data generation using generative AI, evolving both model and infrastructure to meet the training and evaluation needs of Zoox’s autonomous driving solution.
• Lead the development of generative models from small scale objects to complete scenarios, from research all the way to deployment.
• Design effective model architectures and sophisticated training techniques, leveraging all the inputs from our sensor stack and the overall large scale data we have at Zoox.
• Collaborate with perception, planning, safety, simulation, and systems teams to integrate your models into our offline pipelines.
• Validate and optimize your solutions using real-world driving scenarios, directly contributing to the safety and reliability of Zoox's autonomous system

Qualifications:

• MS or PhD in Computer Science, Machine Learning, or related technical field
• Demonstrated experience architecting, training and deploying large models such as diffusion, flow matching, GANs and/or NeRFs.
• Experience building and maintaining ML training pipelines, including data preprocessing, model training, and evaluation
• Proficiency in Python and ML libraries (PyTorch, NumPy) demonstrated through professional or research projects
• Experience training with large scale datasets (e.g. tens of millions of videos)

Bonus Qualifications:

• Publications in top-tier conferences (CVPR, ICCV, RSS, ICRA)
• Experience with autonomous robotics systems
• Experience implementing 4D Gaussian Splatting