Led the analysis efforts for improving the speed and reliability of the vehicle start-up process, which involved cross-functional collaboration with stakeholders from operations, hardware, and autonomy teams

Architected and implemented an Extract, Transform, Load (ETL) pipeline that processes 1M+ raw vehicle data daily from the fleet through Google Pub/Sub into reliability and performance metrics

Created and deployed a fault-injection framework to scalably test 70K+ failure modes across the autonomy stack in the simulation environment, instrumental in risk assessment and mitigation

Productionized and maintained a framework to enable runtime performance analysis of autonomy modules in cloud-based simulation, achieving a 50% reduction in turnaround time through process parallelization

Revamped the weekly onboard systems report, cutting preparation time by 60% with Kubernetes and ReTool for enhanced process efficiency

Improved the Test & Learn software’s flexibility in matching test and control groups and added a new analysis method in the test simulator

Implemented usability improvements including front-end redesign for the test simulator in React and TypeScript

Set up a workflow that uses Splunk to query for persistent unit test failures in Jenkins logs and notifies teams through Slack

Engineered a gRPC-based microservice to streamline Virtual Machine (VM) backup management, facilitating asynchronous backup scheduling among VM hypervisors and the central backup service

Revised the backup process by implementing a new workflow that clones VM instances to images, improving data integrity compared to traditional immutable snapshots

Assisted in teaching CS 250: Computer Architecture and CS 230: Discrete Mathematics for CS

Duties include: grading homework assignments and exams; leading out of class recitation sections with 25 students; holding office hours; preparing study materials; responding to student questions

Topics covered include: C, assembly (MIPS), processor design, memory and I/O

Invited more than 5 academics and industry experts to serve as guest panelists for the annual summit, which focused on opportunities and challenges in the age of big data

Continued maintenance and improvement of the summit website, which was developed with Bootstrap framework and deployed on Netlify platform

Devised a novel approach to synthesize a group of related compounds with AIE properties and characterized them using UV-Vis and IR spectroscopy

Investigated the prospects of developing such compounds for trace explosive detection and fabricated a film-based prototype to evaluate the efficacy of such applications

Co-authored and published WIP: The Cyber-Physical Immune System in Proceedings of the 2021 International Conference on Embedded Software (EMSOFT ’21)

Proposed a conceptual approach to secure Cyber-Physical Systems (CPSs) by deploying an independent network of monitoring components that adapts to the changing environment and reports threats or anomalies

Validated CPIS in the case of a vehicular cruise control system by developing a simulation of the interaction between the vehicular controllers and the physical environment

Developed a system that enables multiple edge-connected devices to collaborate on image recognition tasks using Google’s ARCore and Firebase platforms

Implemented a module that determines whether the fields-of-view of multiple cameras collide using ARCore’s CloudAnchors

Investigated the spatial and temporal correlations in the images captured by devices from different positions and of heterogeneous quality to improve recognition accuracy