Kuiper provides internet for the masses via LEO satellites

* Led the design and implementation of the customer terminal demand aggregator service in Go, which will serve 400 million customers

* Provided technical leadership for back-end development

* Created hierarchical state machine based mode manager for Kuiper satellites, ensuring ~100 sub-systems to be synchronized and run in the expected order

* Improved satellite mode manager state machine's error handling, sped up self-recovery from ~1 hour to ~45 seconds

* Designed and implemented fault injections in end-to-end tests with optional HIL

* Leads a team of 5: designer, front-end developer, embedded engineer, and video processing engineer.

* Manage hardware acquisition and business relationships with 4 companies.

* Created patent-pending VPN and packet-aware Scheduler in Rust to create a low-latency (2 microseconds data processing) and durable bi-directional network connection between vehicle and remote driver console.

* Architected and implemented a GPU-accelerated Media Pipeline with 6 camera sources. With an ultra-low bandwidth requirement (1.6 Mbps) and low end-to-end latency (100 ms) which delivers good video quality (80 VMAF score).

* Created WebRTC P2P network connection in C++ from remote driver console to vehicle with custom congestion control that utilizes TWCC and proprietary VPN metrics to achieve ultra-low retransmissions. In 4 hours of driving within the city of Las Vegas, there were 5 NACKs (retransmissions) and zero freezes.

* Created remote driver console UI in React including authentication/authorization, pick-up/drop-off scheduling queue, and notifications.

* Created a ROS Node in C using the CAN protocol to interface with vehicle’s essential functions: throttle, brake, steering wheel angle, turn signal, door lock, gear selection, emergency signal. Which is transmitted and controllable by remote driver console.

* Created a user-space driver with Linux evdev in pure Go for remote driver console to interface with vehicle’s essential functions while providing force-feedback adjustments.

* Reverse engineered Kia Niro 2020 EV using an oscilloscope to reverse raw voltage signals, enabling remote control of the vehicle.

* Created a customer-facing progressive web app (TailwindCSS, Typescript, React, NextJS) to create and manage reservations, driver license verification, authentication/authorization, payment, scheduling, and availabilities leveraging serverless task-scheduling and custom UI components

* Created a high-performance rosbag parser (under 1 ms for 10 GB of data) in pure Go allowing data analytics in the cloud without ROS dependencies.

* Created a data logger that extracts metadata from vehicle such as drive distance, average/min/max speeds, etc, then upload the data to the cloud for archival and big data query.

* Created Node.js ROS Node in vehicle to allow customers to send navigation from the Halo app to the vehicle’s in-car display

* Created a benchmark tool for objectively determining video encoding quality and performance.

* Created GPS interactive map which visualizes network performance based on RTT, temperatures, and NACK counts.

* Created Node.js ROS Node to listen to vehicle’s telemetry data such as GPS, temperature, and battery level. Which is stored in Google Cloud Database and shown in remote driver console, in-car-display, and customer-facing app.