Dylan Hunn

linked.com/in/dylhunn github.com/dylhunn twitter.com/dylhunn

Hello! I'm a software engineer in San Francisco and a Member of Technical Staff at OpenAI. I believe that we should build artificial general intelligence safely, and distribute the benefits equitably.

Previously, I worked on the Angular web framework at Google, especially on Angular's template compiler.

Beyond my work, I'm an avid snowboarder, science fiction fan, and board gamer.

Click for my full résumé.

Work Experience
OpenAI San Francisco Member of Technical Staff October 2024 — present
Building experimentation systems and signal-collection methods to improve our models.
Google San Francisco Software Engineer July 2018 — September 2024
Worked on Angular, a major open-source web framework, primarily focused on the Angular compiler.
  • Lead author of Angular’s new template compiler (“Template Pipeline”), which transpiles user templates written in Angular’s custom template language into Javascript that renders and updates the application.
  • Built a custom IR and implemented new optimizers, reducing user bundle sizes up to 5%. Enabled new features such as default content projection.
  • Built a compiler-powered import graph analysis feature, allowing the Angular language service to automatically import components when their selectors are used in a template.
Previously on Plaque, a distributed graph programming framework used for training large ML models.
  • Built a distributed tracing system designed to support extremely large graph executions. This was used for profiling in the Pathways ML framework, including for PaLM training.
  • Built an in-browser profiler for inspecting very large Plaque traces, with the ability to automatically fetch traces from individual participant nodes.
YouTube Mountain View Software Engineer Intern Summers 2015 & 2016
Worked on the YouTube Mix team twice as an intern, building the backend that powers YouTube Music in-app auto-playlists. Implemented variety filtering for the YouTube Music app, using a new embeddings model for the music corpus. Grew YouTube Radio watchtime by over four percent.
Education
Stanford University Palo Alto September 2014 — June 2018
Bachelor of Arts and Sciences, Computer Science and Music (joint degree)
Teaching Experience
Stanford University Palo Alto Section Leader January 2015 — June 2017
Section Leader for CS 106, Stanford's family of introductory computer science courses.
  • CS 106A: Programming Methodology (Winter & Summer 2015, Winter 2016)
  • CS 106B/X: Programming Abstractions (Spring & Fall 2015, Spring 2016, Spring 2017)