As a Data Scientist on the Behavior Evaluation team, you will be the statistical anchor ensuring our autonomous driving systems navigate highway environments with world-class safety, efficiency, and comfort. Highway evaluation presents a unique industry challenge: verifying vehicle behavior at high velocities where the margin for error is razor-thin, and critical edge cases are buried in petabytes of data.
In this role, you will bridge advanced statistical methodology with scalable software engineering. You will design the mathematical frameworks, statistical tests, and data-driven metrics that evaluate our planner's decisions. Working directly with large-scale simulation and real-world fleet data, your insights will define our validation pipelines, identify behavioral regressions, and directly shape the software powering our next-generation autonomous fleet.
Design Advanced Experimental Frameworks: Formulate robust statistical models, hypothesis testing frameworks, and quasi-experimental designs (such as synthetic controls or matching) to rigorously validate highway planner behavior in simulation and shadow-mode deployments.
Model Tail Risks & Rare Events: Use Surrogate Safety Measures (e.g., TTC, PET) to accurately model and predict low-frequency, high-severity edge cases that traditional mean-based statistics miss.
Architect Scenario-Based Metrics: Own and mature critical behavioral KPIs, utilizing data stratification to analyze complex driving scenarios (e.g., high-speed merging, cut-ins) while proactively identifying statistical anomalies like Simpson’s Paradox.
Surface Statistical Edge Cases: Apply data mining and advanced statistical techniques to isolate low-frequency, high-severity edge cases and systemic Autonomy engineering debt.
Education: Bachelor’s or Master’s degree in a highly quantitative field (e.g., Statistics, Mathematics, Data Science, Operations Research, or a related field with a strong statistical focus).
Experience: 3–6+ years of professional experience as a Data Scientist or Quantitative Engineer, with a proven track record of landing data-driven impact.
Strong Statistical Foundations: Deep understanding of hypothesis testing, experimental design, regression analysis, non-parametric/resampling methods (e.g., bootstrapping, permutation tests), and time-series analysis handling autocorrelated data.
Strong Programming: High proficiency in Python (Pandas, NumPy, SciPy, scikit-learn) and the ability to write highly complex, optimized SQL queries for massive distributed databases.
Robotics or Autonomy Background: Experience analyzing spatial-temporal data, sensor logs, or vehicle telemetry from robotics, autonomous vehicles, or aviation systems.
Simulation-Based Testing: Familiarity with validating software systems using empty-world or simulation platforms at scale.