About Me

Not your traditional developer, I graduated with a double-major in Economics and Chinese in 2013 from the University of Virginia. The following year, I graduated with a Graduate Certificate in International Relations from the Hopkins-Nanjing Center in Nanjing, China. After graduation, with my Chinese language needs spurring me on, I began studying iOS development with the goal of building a Chinese language study application. After nearly 7 months of self-study, I officially began my development career by publishing my Chinese language app Earlyworm on the App Store. Shortly thereafter I began working full time as an iOS developer in Shanghai, China.

In 2015, I joined DayDayCook and was quickly promoted to iOS team lead. For one and a half years I have led iOS development and seen our app featured on the front page of the Chinese, Hong Kong, and Taiwanese App Stores over seven times each. I have worked closely with the App Stores to develop applications for new platforms like Apple TV, Apple Watch, and iMessage. Our iPhone and iPad applications were named to the Hong Kong App Store’s “Top Ten Apps of 2016” list, and this year DayDayCook was named one of the top 100 startups in China by the Chinese tech magazine 创业邦. Over the past two years our applications have been downloaded millions of times by users in over 30 different countries. Along the way, our company raised a combined $30 million in funding from investors like Alibaba, 500 Startups, K11, and MFund, and IPO’ed on the NASDAQ exchange with the ticker DDC in 2023.

In 2018, I moved to San Francisco to join Meta and work on AR/VR development. Since joining the company I have worked on three separate teams ranging from operating system development to end-to-end ad delivery on newsfeed, and finally building internal tools using LLMs to automate code-writing.

I have been studying neural networks and machine learning in my free time since 2015 and have recently had the opportunity to apply them in my language learning application Earlyworm. The application utilizes categorization, named entity recognition, clustering, summarization, recommendation, and vector search over embeddings. I made extensive use of LLMs and machine learning libraries, including fine-tuning several text-davinci-003 models and building my own clustering, summarization, and recommendation systems.

Recently I’ve been obsessed with using LLMs to fully automate code-writing. I’m inspired by work like Trace and tldraw but firmly believe the entire development process can be automated, unlocking a new role for software engineers that is freed from the labor of drafting code and more focused on the critical thinking around which features to add and how to improve existing systems.