About the job
The Predictive Planning team (PrePlan) develops and deploys state-of-the-art machine learning solutions that predict the future state of the world and plan the Waymo Driver’s behavior. Our mission is to transform Waymo's unprecedented scale of driving data into robust, generalizable, and performant deep neural networks. These models enable the autonomous vehicle to navigate complex environments safely and efficiently.
Responsibilities
Develop the next-generation ML-powered prediction and planning system to enhance the performance and capabilities of the ML driver and support the rapid scaling of Waymo’s business.
Frame open-ended, real-world challenges as well-defined ML problems; research, develop, and apply cutting-edge ML techniques, including foundation models and reinforcement learning, for the planning and prediction tasks of autonomous vehicles.
Collaborate with world-class researchers, engineers and product owners to create safe, smooth planning behaviors for all road users and to meet product requirements.
Develop and evaluate large models, and integrate them into Waymo’s production planning software for real-world applications through close partnership with the Planner and Research teams.
Qualifications
Minimum
BS in Computer Science, ML, Robotics, similar technical field of study
2+ years of experience in Machine Learning modeling and/or Autonomous Vehicles
Demonstrated contributions to the ML community through publications, open-source projects, or significant industry impact
Hands-on experience with modern deep learning libraries (eg: TensorFlow, JAX, Pytorch)
Proficient programming skills (eg: Python, C/C++)
Strong analytical and debugging skills
Preferred
MS or PhD in Computer Science, Machine Learning, Robotics, or a related field
Publications in top-tier conferences such as ICML, NeurIPS, CVPR, ICCV, ECCV, ICLR, IROS, CoRL, ACL, or EMNLP
General software engineering experience solving motion planning or related robotics problems
Experience applying or evaluating ML-based systems in production environments
Experience with performance optimization of deep models, including with respect to specific hardware architectures