Built the $400 Co-pilot® for working drivers (https://youtu.be/gxtIqNmpFek)Bluebox® Co-pilot® is the only product in the world capable of curating training data needed for vision-only self-driving AI model with cars already on the road (plug & play with any cars, 5 min install) while also adding 24/7 around-car security with 4G live stream & life-saving ADAS features (e.g. collision warning) for 1.4B drivers whose only option to get ADAS today is to buy a new carDemo video of Co-pilot® (installed in old Prius) driving me autonomously from Seattle to SF: https://youtu.be/t7xdzmf1P84Currently, Co-pilot® is running the openpilot model (using less than 10% of inference compute available). Vehicle control will be disabled for most drivers & same AI model will be used for ADAS (e.g. collision warning)https://bluebox.io

Developed a full upper-body wearable system to provide haptic feedback solution to touch and feel virtual reality. Responsibilities included system architecture development, PCB design/assembly, body tracking through inertial measurement units, serial and wireless communications, battery and power management system, firmware development, and business development with manufacturers and suppliers.Began as a college project for Intel Cornell Cup and Microsoft Imagine Cup until the unexpected acceptance to the 1st batch of Y Combinator Fellowship (raised $500k angel round). Valuable learning experience that provided many lessons in building a defensible/scalable hardware company (don't build hardware unless it's absolutely necessary to enable new category of software).

I implemented autonomous mobile robot navigation (same traditional stack used by Waymo & co. to burn cash for no reason) and researched more efficient boundary box optimization for obstacle avoidance. Realized that the traditional approach will never work in the real world. Advised by Professor Thomas Howard (Navigation @ 2007 DARPA challenge CMU team)I also began designing and building a research quadcopter platform from scratch for the lab - would've charged by landing on inductive charging pad with guiding structure for easier landing.Didn't have enough time to finish the project before graduating but left design and documentation for another assistant to finish.

Research fellowship at the Robotics and Artificial Intelligence Laboratory (RAIL) advised by Professor Thomas Howard.Worked on developing an adaptive trajectory optimization software and efficient collision detection object models for the TurtleBot.

Assisted with various IT services including IT troubleshooting and computer repair.Acquired fundamental knowledge in network security.