Houjun Liu
I'm Houjun, and I'm doing my BSCS and MSCS at Stanford. My friends call me Jack, and you can too if you'd like :)

If you want to say hi, I can be easily reached via the electronic mail service.

Jack does things

I’m currently making GPUs go brrr at SISL advised by Mykel Kochenderfer and StanfordNLP working with Chris Manning. I’m also a research software engineer at TalkBank lead by Brian MacWhinney.

I don’t really have a good answer of what I work on and lowkey its kind of jit compiled based on what part of NLP or RL I happen to be thinking about. (Some) current threads include…

  • Thoughtbubbles: where we make some architecture shenanigans to induce adaptive computation from an LLM from just unsupervised pretraining data
  • ASTRA: where Schmidt Science foundation graciously handed some funding to us to make LMs safer, and we built a package out of it using a method I developed called ASTPrompter
  • Stanza: occasionally I larp as someone that still work on classical computational linguistics and bake a model for coreferences or constituencies
  • Batchalign: I maintain an speech-recognition swiss army knife package that a bunch of clinicians around the world use for speech sample analysis

Previously I made a task management app with some friends, and worked at a computer vision startup called Dragonfruit AI. I still daydream about being an indie developer of productivity apps and so am currently making another one.

Jack takes a lot of notes

I have a crappy memory but I like to learn a lot so I spent an inordinate amount of time writing notes, some of which you can find here ☞

You can also use the search bar on the upper right to find some more.

Friends?

Contrary to popular belief I spend a bunch of time outside and have actual, human friends. You should meet them, they are very cool.

  • Sanity (David, Hux, Alb, and Zach) is a friend group formed over COVID that I still consider my best friends
  • I feel very lucky to have met many legendary friends during my time at Stanford, including Joseph, Mason, Danny and John; some are so legendary they don’t let me list them here mmmm
  • I’m privileged to have enjoyed a long and productive friendship with two close collaborators, Amelia and Robert, and similarly I’m very grateful to friends and colleges at SISL and Stanford NLP

Picture?

If you want photographic evidence of my face, here you go. I also spend a decent amount of time these days frolicking in the outdoors, which tends to generate pictoors which you can view here.


Last ModifiedPage
January 14, 2026SU-SOC175 JAN122025
January 14, 2026SU-PHIL2 APR012025
January 14, 2026SU-EE364A JAN132026
January 14, 2026Stanford UG Courses Index
January 14, 2026Reliable RL
January 14, 2026principles of biomedical ethics
January 14, 2026Preference Elicitation
January 14, 2026operations that preserve fuction convexity
January 14, 2026norm
January 14, 2026model-free reinforcement learning
January 14, 2026model-free inte
January 14, 2026Medical Ethics Index
January 14, 2026Jensen's Inequality
January 14, 2026function convexity conditions
January 14, 2026expectation maximization
January 14, 2026Euclidian Geometry Crash Course
January 14, 2026Convex Optimization Index
January 14, 2026convex function
January 14, 2026constructive convexity verification
January 14, 2026China Economy Index
January 9, 2026perspective function
January 9, 2026extended-value extension
January 8, 2026SU-EE364A JAN082026
January 8, 2026supporting hyperplane theorem
January 8, 2026SU-STS115 JAN072025
January 8, 2026SU-SOC175 JAN072025
January 8, 2026SU-EE364a JAN062026
January 8, 2026seperating hyperplane theorem
January 8, 2026log determinant
January 8, 2026convexity preserve line restriction
January 8, 2026convex set
January 6, 2026virtue ethics
January 6, 2026SU-STS115 JAN052025
January 6, 2026SU-SOC175 JAN052025
January 6, 2026quadratic programming
January 6, 2026optimization (programming languages)
January 6, 2026optimization (math)
January 6, 2026non-linear optimization
January 6, 2026MOEReview Yun: Inference-Optimal MoEs
January 6, 2026knowledgebase testing page
January 6, 2026Deontology
January 6, 2026convex optimization
January 6, 2026compiler
January 6, 2026AI Incidents Policy
December 25, 2025stuff
December 19, 2025MOEReview Zhang: Mixure of Attention Heads
December 19, 2025MOEReview Tan: Scattered MoE
December 19, 2025MOEReview Sukhbaatar: Branch-Train-MiX
December 19, 2025MOEReview Shen: ModuleFormer
December 19, 2025MOEReview Sharma: LAZER