Within Perception, the Data Primary team builds and operates the data infrastructure that powers Waymo's perception foundation models. We design scalable pipelines that process petabytes of sensor data and allow everything from rare event mining to large-scale offline model deployment. Our work directly supports model training, evaluation, and production, ensuring that perception models have the high-quality, diverse data they need to perform safely. We collaborate across ML, Simulation, and Infrastructure teams to integrate data into the model lifecycle and help bring new capabilities to the Waymo Driver.

In this hybrid role, you will report to an engineering manager.

You will:

• Define the technical strategy and long-term architecture for petabyte-scale data pipelines that power Waymo’s perception foundation models.
• Drive cross-functional initiatives across ML, Simulation, and Infrastructure teams to solve complex data ambiguity and integrate data into the model lifecycle.
• Architect advanced systems for data mining, curation, and active learning using ML techniques (e.g., embeddings, vector search, offboard inference).
• Serve as a technical lead and mentor for senior engineers, setting engineering standards and conducting code/design reviews for C++ and Python systems.
• Lead the design evolution of critical data infrastructure to ensure high scalability and reliability for offline model deployment and rare event mining.
• Accelerate model deployment velocity across new cities and vehicle platforms by optimizing end-to-end data lifecycles.

You have:

• C++ programming skills (required), with Python experience.
• 8+ years of experience with large-scale distributed data systems (e.g., Spark, Beam, Dataflow), or PhD and 5+ years of relevant experience.
• Proven track record of architecting and delivering complex, multi-system ML data engineering or active learning solutions.
• Experience providing technical leadership, managing project priorities, and influencing technical strategy across multiple teams.
• Deep understanding of end-to-end ML data pipelines and their interaction with model training and evaluation.

We prefer: