• Vehicle & robot sensor integration with big data platform developement, analysis, and performance optimization

• Sensor software development & data analysis for GPS, IMU, positioning, etc. for enhanced navigation and Autopilot features

MSD (Momenta Self-Driving) -> UNP (Urban Navigation Pilot) -> Motion Planning Team• Deployed & optimized Constrained Iterative LQR Algorithm for lateral motion planning with added safety constraints that improve urban driving simulation success rate by 10%• Designed lateral safety checker for bad planned trajectories• Optimized build process for CiLQR docker to be 50% faster

• Led the team in building UC Berkeley’s first autonomous electric formula race car• Implemented Key-point Regression / PnP algorithm for Monocular camera distance estimation• Integrated Min-Laptime trajectory planner into AMZ simulator for optimal race-lane demonstration and control testing• Secured 10,000+ USD V-LiDAR sponsorship for Gen-2 autonomous system

• Designed fast and accurate perception pipeline with monocular camera set-up using Key-point regression, CNN, and PnP algorithm• Programmed Lap-time optimization trajectory planner and N-MPC path follower using IPOPT/CasADi

• Debugged CARLA simulator on Docker to fix traffic management tools & programmed traffic simulation with traffic lights/stop signs• Implemented an LSTM w/recursive multi-step strategy for trajectory prediction & achieved a <0.5 m RMSE for 10s forecast

• Implemented 1588 PTP using C++ that improved Lidar/camera synchronization accuracy by 99.1% from 46 to 0.4 millisecond error• Developed dynamic delay and automatic timestamp matching software package on Python to improve sensor fusion accuracy by ~40%• Led the research in improving multi-Lidar point cloud clustering/matching using K-means/NDT algorithm via Python on ROS• Designed a low-latency energy-efficient PTP switch w/ RFQ written that reduced the total build cost by ~80% from $600 to $130/set

• Program managed 9 Model 3/Y 2170 battery programs using Jira & Confluence that would save Tesla 10 million USD per year • Cooperated with multiple teams to execute a new battery project and construct a ~420k USD trial manufacturing line within 6 weeks• Designed a benchmark test for battery module width control that improved time efficiency by ~85% from 2 hrs to 20 min per trial• Redesigned the structure of battery interstitial strips with full-surface laminated PE layer to improve the tensile strength by 60% • Designed a squared cross section contoured lock pin for coolant jumpers using CATIA to increase slip-off safety margin by 20% • Responsible for solving an urgent battery launch issue using Python + Excel data analysis to create a heat map of failed cells that validated a risky change action plan which eventually saved 2000+ hold cars and 7000+ battery packs from recall/remanufacture

• Developed a multi-sensor perception&control algorithm for robot racecars w/camera + infrared sensors powered by a STM32 via C++• Programmed IMU/DDE modules in addition to GNSS/RTK on Jetson AGX Xavier via C++ to improve navigation accuracy by 25% • Co-led algorithm development & data analysis for GNSS spoofing detection under strong interference using auxiliary sensor system

• Co-led the design&testing of the 205 mm TC project for <200 Nm vehicles that would boost Valeo’s annual profit by 1.5 million USD• Redesigned turbine blade w/CATIA & performed FEA analysis w/ANSYS for max cost-down while maintaining structural integrity

• Solved for stable boundary conditions for a helical rod structure with discrete structure analysis using Matlab/C++• Implemented Hessian & Jacobian matrix for energy analysis and finite element method for force/motion analysis • Found the correlation between the helical rod’s critical angle of twist and rod pitch, radius, etc. using Matlab/C++

• Participated in designing the actuator mechanism of the 3rd-Gen E-CVT that keeps engine RPM within max power band• Analyzed thermal performance of the transmission system and proposed several designs of cooling systems for better longevity • Improved power transmission ratio by 8% by programming the actuator arm’s engagement logic with data analyzed using Matlab• Reached out to 20+ local tech companies through emails/calls for professional technical assistance and manufacturing resources