I'm building Friendshift, where I'm working on humane technology for the friendships that matter. Before this: seven years at Google, building ChromeOS, and most recently Aluminium OS, the Android-based platform behind the new Googlebook. Before that: UCLA, internships at Uber, Google, and Intel, a CubeSat, community radio, college radio, and newspaper.
The case is well-documented: in-person time with friends has fallen 57% in the last decade, with measurable costs to health, meaning, and the texture of ordinary life. We're building for the friend who wants to be a better friend, without that work feeling like one more thing to manage.
My role is broader than any single function. Engineering is the strongest thread — I own the architecture from device through cloud and lead the AI work — but the job lives just as much in product, design, and the thousand small judgment calls that decide whether a thing feels human or not. The timing is what makes it possible: foundation models can finally hold the kind of ambient context a thoughtful friend would hold in their head, without that context becoming ad inventory. Friendshift is what that looks like when you point the technology at friendship instead of attention. It's the most demanding work I've done, and the work I've been preparing for since I was a kid in Austin building things on the side.
There's a name for the relationships that hold those pieces: chosen family. The term comes from the LGBTQ+ community — the family you build for yourself out of friends, partners, and mentors, the people who love and accept you when the family you were born into can't, or won't. For those whose families of origin can't hold them at all, that kind of family is everything — a place to be safe, to be loved as you are.
During COVID, I saw what tending chosen family actually looks like, lived out up close. My partner at the time cared for his friendships with a kind of deliberate attention I hadn't seen anyone give before. Watching him, I realized I wasn't tending mine anywhere close to as well as I wanted to.
I'm someone who can go a long time without talking to a friend and assume we'll pick right back up. That works for a few; it fails for most. My closest people are scattered across cities and fields (a precious few I've known since my early teens), and the work of staying close has gotten harder, not easier. I'm not as good at it as I want to be.
The bet behind Friendshift is that AI is going to give us back time, and a lot of us will realize we've been spending that time everywhere except on the relationships that mattered most. We're building the technology to lower the activation energy of tending. Not to replace the human work, but to make sure it actually happens. For my own scattered friendships, and for everyone else's.
Humane tech for the friendships that matter. Lead engineering and AI; partner with Faust Whale on product, hardware, and direction. Hardware, mobile, cloud, and the connective tissue between them.
Seven years shipping software end-to-end across the Chromebook stack and Aluminium OS, its Android-based successor powering the new Googlebook lineup — from the user-facing Settings screens down to the C++ system services that manage cellular and network connectivity on the device. Earned OWNER on five Chromium modules spanning backend, middleware, frontend, and the end-to-end test suite — full-stack review authority that's rarely granted, and earned over years.
Cellular Chromebooks for public-school students. I was a core engineer on the team that delivered this end-to-end — building the admin console, the backend, and parts of the on-device install — so districts like NYCDOE and LAUSD could give every student a Chromebook with its own cellular connection, and kids without reliable home internet could still get online and do their homework.
Beyond that: was a core engineer on the rebuild of how Chromebooks discover and join cellular networks, across UI, backend, and device, including the migration that moved existing users onto the new system. Core contributor to Phone Hub — the feature that puts your phone's notifications, recent browser tabs, and hotspot one tap away on your Chromebook — where I shipped the auto-tethering integration, the recent-tabs surface, and tooling that the rest of the team used to ship their pieces faster.
The through-line across all of it: assistive technology. Every settings flow, every Phone Hub surface, every cellular setup screen I shipped was built to work for screen-reader users — and to render correctly in right-to-left languages like Arabic and Hebrew. Done right, accessibility isn't a compliance checklist; it's the discipline of building software that meets every kind of person where they are. Friendshift is built on the same instinct — assistive technology for a different problem: navigating adult life without losing the people who matter.
Took the Computer Engineering track of my EE degree — the seam between hardware and software. Senior project: a real-time motion classifier — a neural network plus a custom peak-detection algorithm, prototyped in Python and shipped as hand-optimized C on a small Intel embedded board with a 9-axis motion sensor, classifying walking, running, turning, stair-climbing, and hopping. Spent years on the ELFIN CubeSat mission — wrote hardware-level circuits (Verilog on FPGAs) for the satellite's payload instruments: a math unit for processing sensor readings, designed to use less of the chip's tightly-budgeted area, plus the data-integrity pipeline that carries readings between the satellite's onboard computers.
Outside the classroom, I shipped real things for the student orgs I was part of — UCLA Radio above all. As DJs we had no working site for listening (the stream lived on TuneIn) and no place on the official site to describe our shows. So I proposed and led the full-stack rebuild — Node.js and Express, the hyped new stack at the time — and added an anonymous live chat so listeners could message whoever was on-air in real time. Through that chat, listeners introduced me to rare electronic tracks I still keep on vinyl today, and I realized for the first time that I wanted to build technology that improves connectivity in all its forms.
Two halves of an engineer were formed in parallel here. Academically — deep in hardware with my fellow EE students through projects and labs — I learned to write software that talks to hardware. The other half — software that talks to people — I picked up alongside the humanities majors I worked with at the newspaper, at UCLA Radio, and later at Berlin Community Radio, by learning how they think. The technical apprenticeship for that side came from CS majors I'd never built with before: my EE classes had been embedded-level (mostly C), not the databases and classical algorithms web development calls for. I learned those on the college newspaper's web team and in a paid iOS developer job at UCLA Library (which also helped pay off student loans). I traded academic quarters for industry along the way — six months at Intel, then back to school, then a summer at Google straight into a fall at Uber — and closed out the degree at Freie Universität Berlin (humanities coursework), staying in Germany on Berlin Community Radio until my scheduled return to California to start at Google full-time.
Built a real-time data visualization for the Uber Driver iOS app in Swift — a demand-forecasting graph rendering historical and live marketplace data so drivers could position themselves more profitably. My first time owning a piece of UI at scale that real people relied on for income. Uber wasn't a startup anymore by the time I got there, but it still had the fast-paced energy of one, and I loved it. I learned a lot about what really goes into shipping mobile — iOS in particular — and how much of the job is coordinating with the engineer on the other platform to keep features and behavior in parity.
Wrote C++ APIs for ChromeOS cross-device communication and parts of the synchronization layer. I'd come in trained on low-level C from school and found a real sweet spot in modern C++ — more expressive, but with the explicit control I liked. Working on Chromium was its own reward: contributing to open source on a codebase real people relied on. But what I remember most is how much better an engineer I became over those three months — and how much of that came from the people I was lucky to learn from. Good enough that I came back to the same team full-time after graduation, and stayed on it for years.
Wrote firmware in C for an IoT wearable and used Python to analyze large datasets of user data — skin-temperature sensor readings streamed off wrist-worn devices — and to automate power-consumption testing and transfer-function verification. Six months of learning what software looks like when bytes have to fight for a battery.
Helped design, assemble, and test prototypes of a feedback-controlled portable cooling and heating device that conforms to the body — my first taste of building hardware that would actually touch a person, and of how much the engineering matters when it does.
Started in visual arts at McCallum Fine Arts Academy for a year in high school, then transferred to Liberal Arts & Science Academy to pursue math and science, interests I'd always held alongside the arts. Engineering became the synthesis: it let me bring the creative thinking fundamental to fine arts into math and science. ("Starving artist" sounded less like a phase and more like a forecast, too.)
I still paint, sketch, draw, doodle, and shoot film on a Contax T2. Making things by hand pulls me in the same delightful way coding does. In the age of AI, coding itself increasingly feels artisanal. That same eye obsesses over pixel-perfect UI.
Off-screen, I gravitate toward Issey Miyake with matters of dress, and you'll find that I have a soft spot for Bauhaus architecture. I'm glad my algorithm keeps a passive pulse on design trends in all their forms.
Mandarin is my mother tongue, and I'm fluent in written Chinese (trad) surprisingly. Half my family is in Kaohsiung; I spent entire summers there growing up and still go back every year. Taiwan is as much home as Austin.
Then German, seven years of it from middle school through college, plus humanities coursework at Freie Universität Berlin, and a fluency I'd generously call "conversational." Berlin also feels like home. I continued living there after my studies until starting work full-time back in LA. I find time to go back every year.
Started on UCLA Radio with carefully curated sets played track by track, learning to blend them over three years of hosting there. Berlin Community Radio came next, where I went on the air through the summer of 2018 whenever I had a new set worth sharing.
Vinyl came with me when those years ended and stayed, the tactile delight I kept when the rest of life moved to screens. My collection runs broad, with a real weakness for rare dance records that have no online trace.
Nowadays I mix mostly at home. My setup includes a pair of turntables and CDJs, linked with a Xone:92 (Allen & Heath). Some sessions are pure vinyl, others pure USB, depending on the mood. Nerd alert! The same parallel shows up in code. Mixing on vinyl is like writing in C: technical, exacting, every beat matched by hand. Mixing on CDJs is like Python or AI: you can hit sync to offload the beatmatching, freeing attention for higher-level creative choices. The skill is moving fluidly between the two inside a single mix while the flow stays unbroken.
Most sessions vanish the moment they end, and a lot of what I love about mixing is exactly that: the unrepeatable shape of a given night. Every so often I'll plan one and record it. Those recordings let me revisit tracks I was deep into at a particular moment, like 25 May, Austin Texas or Living Room Booth 1.
The needle finds the groove and the room changes, a small physical delight that's on my mind every time we make a hardware decision for Friendshift.