About Me

I'm a programming languages researcher and software engineer, currently at Google. There, I work on the AI Edge Performance team on YNNPACK and Slinky.

I am also a co-lead maintainer of the Halide language. I am responsible for the build system, packaging, testing, and release processes. I have also been researching new ways to manage quantized kernel variants and have been adding features to the x86, ARM, Metal, and CUDA backends to support ML inference in Halide.

I got my Ph.D. in 2022 from UC Berkeley, advised by Jonathan Ragan-Kelley. I focused on building and formally analyzing user-schedulable languages for high-performance computing. Before Google, I worked on Halide research and engineering at Adobe, and before that I spent a year after my Ph.D. at Qualcomm AI Research optimizing LLMs for edge devices.

My recent papers include:

In 2017-2018, I rewrote the compiler for the P programming language.

In the past, I interned with the Intune team at Microsoft, the Facebook AI and Research team, and Microsoft Research. My undergraduate thesis was on software synthesis; my CAV 2015 paper "A Type-Directed Approach to Program Repair" describes a tool called "Winston" that autocompletes and repairs Java expressions (using classical, pre-LLM, techniques).

In addition to my academic interests, I play progressive metal music on the seven-string electric guitar, and have spent the last few years learning how to cook modernist cuisine.