Work Experience

Generated by Veo 3

Data Scientist, Research | Mar. 2025 – present

Senior Data Scientist | Aug. 2023 – Mar. 2025

Developed metrics to measure autonomous vehicle safety, and conducted comprehensive safety analysis with cross-functional teams to support the launch decision for new markets.
Leveraged data from published papers to benchmark Cruise’s safety performance against top competitors, providing quantification of the comparison and identifying key areas for improvement.

PhD Intern, Data Scientist | Jun. 2022 – Sep. 2022

Built data miners and visualizers in SQL and Python that can efficiently filter and describe 1M+ driving scenarios based on kinematics features.
Employed PCA and T-SNE to reduce the dimensionality of the feature space and used multiple comparison techniques to assess the realism of different scenarios.
Trained an anomaly detection model using Isolation Forest to evaluate the realism of driving scenarios in simulation, with analysis results showcased in Cruise’s presentation at KDD ’22.
Helped the simulation team to identify unrealistic scenarios based on the computed realism score.

Statistician Intern | Jun. 2021 – Sep. 2021

Investigated and compared adaptive two-stage design algorithms (Jones, Tournoux-Facon and Parashar designs) that incorporate biomarker status using simulated dataset induced by real clinical trial data. The adaptive two-stage design algorithms could help to identify the targeting population that might benefit from the drug, so that less drug development will be stopped in phase II due to treatment effect dilution.
Enhanced the robustness of the current algorithms by improving the type I error and power calculation, as well as optimizing the search strategy.
Utilized RCpp to write a more efficient searching algorithm and implemented the improved algorithms to an R package and R shiny app.

This work is presented in a poster session at JSM 2022.