Welcome to my homepage.

I work on programming language and compiler technologies for machine learning systems on the Google Brain team. Apart from engineering I enjoy reading bits of theoretical linguistics and doing accents.

Experience

  • Software Engineer, Google Brain, Mountain View, CA, Dec 2017 - Present
    • Part of the core team that launched Swift for TensorFlow, a new programming model for TensorFlow with first-class compiler support.
    • Demo’d Swift for TensorFlow at the TensorFlow Dev Summit 2018 (video), with Chris Lattner.
    • Contributed to all components of the Swift for TensorFlow project. Drove the design and implementation of the core tensor APIs, first-class automatic differentiation, and prototypical high-level APIs.
  • Natural Language Processing & Machine Learning Engineer, Apple Inc., Cupertino, CA, Aug 2017 - Dec 2017
    • Briefly worked on applied natural language processing and localization technologies for Siri.
  • Founder and Architect, The DLVM Project, Dec 2016 - Aug 2017

  • Member, LLVM Compiler Research Group, University of Illinois, Fall 2016 - Aug 2017
    • Designed and implemented DLVM.
    • Inadvertently designed logos for ALLVM and DLVM.
    • Deliberately meowed a lot on IRC.
  • CS 421 Course Assistant, Department of Computer Science, University of Illinois, Spring 2017
    • Designed and led the programming assignement “MP6” - Implementing Scheme in Haskell.
    • Favorite comment from students: “MP6 was my favorite CS assignment of all time. I wish I had more time to do it justice.”
  • Software Engineering Intern, Apple Inc., Cupertino, CA, Summer 2016
    • Contributed to Siri speech crowdsourcing tools and NLP.
  • Software Engineering Intern, Qualcomm Inc., San Jose, CA, Summer 2015
    • Contributed to 802.11ac/ax WLAN firmware.

Contributed Talks

Publications

  1. “A modern compiler framework for neural network DSLs”
  2. “A modern compiler infrastructure for deep learning systems with adjoint code generation in a domain-specific IR”
  3. “DLVM: A modern compiler infrastructure for deep learning systems”
    • Richard Wei, Lane Schwartz, and Vikram Adve
    • Pre-print

Education

  • University of Illinois at Urbana-Champaign, May 2017
    • B.S. Computer Science & Linguistics
    • Bachelor’s thesis titled “DLVM: A Compiler Infrastructure for Deep Learning Applications”
    • I was never a fan of taking classes.

Miscellanea

  • I don’t know any sort of web development.
  • I like to build reliable software from the bottom up.
  • I am a good cook.
  • I am @rxwei on GitHub.
  • I play the Erhu.
  • I backpack once a year.