Senior Staff Data Scientist, TLM

Rigorous behavioral evaluation of the Waymo Driver is a critical part of scaling our ride hailing service and achieving Waymo's ambitious goals. Autonomous driving presents a new paradigm in data science: in addition to leveraging data collected on-road, we generate our own data using state-of-the-art simulation technology-resulting in denser signals and challenging new problems in estimation and experimental design. Our data science team is tasked with providing early insights into the behavioral performance (safety, progress, quality) for new driving software, new hardware platforms, and new territories. We also develop and deploy statistical and testing methodologies to bolster the rigor and actionability of these insights. We work closely with the Onboard engineering team developing the Waymo Driver to ensure they have the evaluation tools to effectively iterate and improve behavior across a wide range of driving contexts. Our work ensures Waymo can achieve targeted behavioral improvements while maintaining the necessary strict safety standard.

In this hybrid role you will report to the DS Lead for Onboard Behavioral Evaluation.

You will:

• Define what a successful behavioral launch looks like for the Waymo Driver by balancing a large portfolio of behavioral metrics covering a diverse range of driving contexts and developing data-driven strategies to evaluate behavioral tradeoffs across these contexts. This eval portfolio and tradeoff strategy will be deployed for use by hundreds of engineers as they iterate and evaluate potential behavioral changes to the Waymo Driver.
• Define and uphold a high bar for evaluation rigor. Be a gatekeeper ensuring that we can confidently rely on the evaluation signals informing behavioral launch decisions. Identify and work to resolve any gaps where current evaluation signals are not providing the necessary actionability for stakeholders.
• Develop and adapt our evaluation framework to accommodate a rapidly evolving collection of operating domains, including new territories and new hardware platforms.
• Leverage behavioral evaluation tools to provide rapid actionable feedback on each new behavioral software release, playing a key role in the release readiness process. Communicate these findings to senior stakeholders.
• Collaborate with leads in product, engineering, and data science to align on roadmaps to unlock key Waymo deployment milestones and collaborate when there are codependencies. Be an opinionated partner informing roadmaps for front-end internal analysis tools and setting requirements for back-end evaluation infra.
• Become an expert in the behavioral software underlying the Waymo Driver, including large-scale ML models, and act as an approver on behavioral software changes.
• Be an active technical contributor on the team, as well as a technical lead and/or manager to several data scientists. Frame and solve ambiguous problems by scoping technical priorities and innovating on statistical methods. Champion data science excellence and provide constructive technical feedback within the team and across Waymo.

You have:

• Degree in a quantitative field (e.g. Statistics, Mathematics, Physics)
• Either a PhD in a quantitative field and 7+ years of industry experience, or 9+ years of industry experience solving data science problems.
• Experience as the technical lead for a broad data science area.
• Expertise using advanced statistical methods in an applied setting; familiarity with ML systems/models
• Demonstrated knowledge of data analysis libraries and packages in Python, R, and/or SQL.

We prefer:

• PhD in a quantitative field
• A demonstrated track record of independently driving data science projects to deliver business value
• Experience solving problems related to Autonomous Driving or Ride Hailing.
• Experience with large-scale evaluation frameworks for software development.
• Experience in adjacent relevant areas like Advanced Machine Learning (Deep Learning and Diffusion models), Traffic Modeling, Safety Evaluation or Prediction