Industry Experience

Rivian

To find close calls, I developed and implemented a safety metric for lateral distance to surrounding objects, including filtering logic to identify relevant vehicles and computing the minimum lateral distance in real-time.
Designed a dynamic Minimum Safe Lateral Distance Threshold model, incorporating reaction time, relative acceleration, and deceleration assumptions to quantify unsafe lateral proximity events.
Conducted A/B testing between a feature branch and baseline, analyzing lateral distance statistics, violations, regressions, and performance improvements to validate system behavior.

To find behavior deficiencies in a self-driving truck’s planning model, I wrote a Python tool for coverage analysis based on internal metrics, which was able to find improvements in simulations created to test the AV behaviors.
Built a detector that achieved >90% precision for scenarios where the AV saw a vehicle on a separated shoulder.

To inform planning and safety mechanism constraints, I formulated and wrote a Python tool to perform a time-interval safety analysis on excessive longitudinal acceleration fault scenarios of a self-driving car.
Derived one dynamic and three static 2D rollover models of a self-driving truck in Python, which served as the groundwork for the safety team to impose motion planning limits that prevent rollover in different conditions.