About

Warren Zhu

Warren 朱富橙

I'm a senior at Harvard University and grew up in Hong Kong and Shenzhen. I want to build tools that help us think through deep and complex problems. I believe this requires building a) adaptive informational interfaces for humans, b) self-organizing tools to manage our knowledge and preferences, c) cognitive environments that feel natural to use.

I love and hate the modern world, and I am especially skeptical of (social) science. I split my time between playing with new technology, reading about old ideas, and being (technology free) with loved ones.

If anything in these webpages provokes any thought, email me about whatever at wzhu@college.harvard.edu! You can title it "Hi" and write anything in the contents :). For more formal details, you can view my resume.

→ Read my latest posts

Education

  • Harvard University (Class of 2026)
    • B.A. in Computer Science
    • B.A. in Statistics
    • M.S. in Statistics

Current Projects

  • I am building LLM-based tools to help people think and articulate their thoughts. Towards this end, I am reading everything I can find about writing. I would also love to talk to anyone about their writing process! (Please email me about this if you have any thoughts or want to chat!)
  • My most recent thoughts are in this post

ML Systems Projects (unfinished! Contact me if you are interested in taking these ideas!)

  • I’ve worked on on the standardization of fault tolerance protocols for distributed ML. “CRAFT: A Pytorch-based Protocol for Composing Fault Tolerance Techniques” is accepted to Pytorch Conference 2025, and I have contributed to torchFT to improve the fault tolerance of distributed ML systems. I have left much undone, and I would love to share my thoughts if anyone is interested in taking them on.
  • I have sketched out how to generalize the PyTorch grid-like DeviceMesh abstraction into a Tree for hierarchical SGD scenarios. I have not implemented it in PyTorch since the engineering work to get it compatible with everything is substantial, but would love to share my thoughts there also.

Statistics Research

I did Harvard’s SPUDS during Summer 2024, sponsored by the great Xiao-Li Meng and under the incredible mentorship of Connor Jerzak. It appeared in clear2025.

Teaching

I had the pleasure to be a teaching assisstant for these wonderful courses:

Fall 2024: Course Assisstant for Stat188 (Variations, Information, and Privacy) under Prof. Xiao-Li Meng, with the Graduate Teaching Assistant Kyla Chasalow. Spring 2025: Course Assisstant for Stat111 under Prof. Joe Blitzstein, and Stat288 under Prof. Xiao-Li Meng.

I have also been tutoring highschoolers after seeing how helpful I could be to my sister in thinking through her academic and life choices. It has been even more fun than teaching college students!

Other Hobbies

I went to Harrow International School Hong Kong, where I had fond memories studying History, Drama, English, Music, and playing Badminton and Rugby, and many, many unhappy memories. I advise that any prospective parents/students to not be fooled by its marketing. From what I can see, it is horribly bureaucratic and unresponsive to student needs. It also appropriates the brand of a British Public School for economic means (which is legal, but I personally think quite unethical). I should note that I respect many great teachers in the school; I wrote the above paragraph out of quite a fit of dissatisfaction with the school.

I had a lot of free time on my hand during COVID and spent a lot of time reading continental philosophy (Heidegger, Hannah Arendt, Hegel, Kant, etc.). I won the Trinity College Philosophy Essay Prize for my essay on Social Media and Hannah Arendt’s Concept of Solitude. I was also highly commended in the Philosophy section of the John Locke Essay Prize for an essay on Heidegger and Morality.

I am a sucker for Wong Kar-wai movies and love Twin Peaks, Brideshead Revisited, and Yes, (Prime) Minister!.

Courses Taken at Harvard

For the interested, here are the technical courses I’ve taken at Harvard, listed in order of influence on me. For Harvard undergrads, feel free to email me about any of these courses!

Bachelors in Computer Science

GPA: 3.92

  • CS226 (Graduate: Distributed Systems): Only full mark Final project in the class on Fault Tolerance in Distributed ML
  • CS145 (Networking at Scale: Audited)
  • MIT6.S982 (Graduate: Deep Learning: Audited)
  • CS143 (Computer Networks)
  • MIT 6.5930 (Graduate: Hardware Architecture for Deep Learning)
  • CS61 (Systems Programming)
  • CS128 (Convex Optimization in Machine Learning)
  • CS2241 (Graduate: Algorithms at the End of the Wire)
  • CS124 (Algo. and Data Structures)
  • CS136 (Economics and Computation)
  • MIT 6.S98s (Graduate: Diffusion: Theory and Practice)

Masters in Statistics

GPA: 3.91

  • Stat220 (Graduate: Bayesian Data Analysis)
  • Stat288 (Graduate: Deep Learning and Earth Observation)
  • Stat188 (Differential Privacy)
  • Stat210 (Graduate: Probability Theory I)
  • Stat212 (Graduate: Probability Theory II)
  • Stat211 (Graduate: Statistical Inference I)
  • Stat242 (Graduate: Time Series)
  • Stat111 (Statistical Inference)

Math

GPA: 3.92

  • Math55a/b (Honors Real/Complex Analysis and Linear Algebra)
  • Math123 (Algebra II: Groups and Rings)

Bad Courses

I have had the pleasure of attending courses (to complete graduation requirements) that I hated to my bones. They helped me appreciate how great my other courses were. Here is the thank you list:

  • CS1060 (Software Engineering with GenAI)
  • Expos20: Is it Okay To Be A Luddite (Compulsory Writing Course)

Posts

subscribe via RSS