Prior work

My recent work in AI research is covered under research.

Earlier, I worked in a few different software engineering disciplines:

Computer vision 2017-2024

While most of my work in computer vision is covered under research, I also worked on applied projects that did not result in publications:

Most recently, I worked on a text-to-image diffusion model trained within Meta's Monetization org to generate ad images.

Before that, I worked on VCE, a video copy detection system that uses a lightweight fingerprint model based on SSCD.

Startup software engineering 2014-2019

I've worked at two startups, joining both at the ~8 person stage. The first, Leanplum, grew to ~200 employees during my time there. The 2nd, GrokStyle, was acquired by Facebook in 2019.

My work focused on backend engineering, but also included sales calls, meeting with customers, writing a javascript SDK, and a bit of mobile development.

SRE and production automation 2005-2014

I began my career at Google in Site Reliability Engineering, which at the time was a new field. I focused on production automation.

A deployment automation system whose development I led starting in 2008 is still widely used within Google. A history of the project was presented at SRECon Americas 2019, and the system is discussed briefly in the SRE book.

Programming languages 2003-2005

I worked as an undergraduate research assistant in a programming languages research group, including on the MetaOCaml language.