Waymo's Release Evaluation org ensures that each version of the Waymo Driver is safe before it hits the road. We build automated pipelines to solve the long tail of rare and exceptional scenarios in autonomous driving, looking for needles in a haystack under both time and resource constraints. Within Release Evaluation, the Sampling and Efficiency team applies importance sampling techniques and machine learning to maximize the statistical efficiency of these discovery pipelines.

You will:

• Develop importance sampling techniques that enable our evaluation pipelines to deliver better signals with fewer resources.
• Find signals in our logs and simulations that might help us to more efficiently discover rare and important events.
• Build systems that systematically optimize multiple objectives under resource constraints.
• Collaborate with other engineers, data scientists, statisticians and the leadership team to deliver evaluation products and help make data driven decisions.
• Champion code health and best practices in a large and complex code base

You have:

• BS in Computer Science, Robotics, Statistics, Physics, Math or another quantitative area
• Fluency with probability and statistics
• Strong self-motivation to navigate complex systems and pursue open-ended problems to completion
• 2-3 years of experience with
  • Navigating and modifying a large code base containing a variety of languages, such as C++, Python and SQL
  • Performing statistical analyses
  • Building data processing pipelines
  • Writing, reviewing, and merging code following industry standards for code health and maintainability

We prefer:

• Experience programming in C++
• Experience developing and evaluating sampling methods
• Experience designing, training, evaluating, and applying ML models
• Experience working in the AV industry
• PhD in a quantitative field