Waymo is looking to improve the rigorous performance evaluation of the Waymo Driver, a critical part of scaling their ride-hailing service and achieving Waymo's goals. This involves developing statistical models, metrics, and measurement frameworks to ensure the Waymo Driver meets strict standards for safety, compliance, and driving/service quality, especially with complex machine learning systems and simulation data.
Requirements
- A solid statistical theory foundation
- Excellent data institution- you are able to quickly grasp data challenges, understand the statistical nuances, propose a strategy to solve those challenges, and implement a solution
- Proficiency in programming with Python with an emphasis on statistical coding, data manipulation, and data visualization; also basic knowledge of SQL
- Experience with developing statistical methodology, casual inference, multiple hypothesis testing, A/B experiments, sampling methods, and/or predictive modeling
- Experience with interpretation and evaluation of complex ML systems and/or foundation models
- Fluency in SQL
Responsibilities
- Create innovative statistical and machine learning methods and practical guidelines for assessing early software versions, prioritizing actionable insights and efficient evaluation
- Develop tools that pinpoint meaningful driving behavior changes resulting from even minor code modifications to major ML model updates, effectively guiding engineer's focus to the signals among noise
- Develop and optimize sim-based eval strategies for a high priority novel problem area
- Validate your methodology and provide code to enable ongoing use of your methodology (by leveraging the Alphabet software development infrastructure)
- Present findings to stakeholders
Other
- Pursuing a PhD in a quantitative field (e.g. Statistics, Mathematics, Physics, Engineering, Economics, Political Science)
- Proven skills in effective communication through presentations and written documentation
- Excellent interpersonal skills
- 3+ years of experience in data wrangling, visualization and data-driven storytelling
- Industry experience
