Jack Cook
Hello! I’m a first-year PhD student at MIT EECS studying machine learning for systems, advised by Song Han. I previously studied at the University of Oxford while supported by a Rhodes Scholarship, and I’ve worked on LLMs at Modal, The New York Times, and NVIDIA.
My research focuses on improving the efficiency of machine learning systems. The brain serves as an existence proof that much more efficient learning is possible than what we see in modern LLMs. While understanding this gap encompasses many different topics, I currently focus on low-precision numerical formats and model architecture design.
Before starting my PhD, I earned master’s degrees in computer science, neuroscience, and social science, and I used to be the director of HackMIT. A long time ago, I was on the founding team of Mixer (formerly Beam), acquired by Microsoft in 2018.
I occasionally blog about my work. If you’d like to be notified when I publish my next post, enter your email below.
Research
Brain-Like Pathways Form in Models with Heterogeneous Experts
Jack Cook, Danyal Akarca, Rui Ponte Costa, Jascha Achterberg
NeurIPS 2025, CCN 2025, UKNC 2025 (Oral)
There’s Always a Bigger Fish: A Clarifying Analysis of a Machine-Learning-Assisted Side-Channel Attack
Jack Cook, Jules Drean, Jonathan Behrens, Mengjia Yan
ISCA 2022
Intel Hardware Security Academic Award, First Place
IEEE Micro Top Picks: Top 12 Computer Architecture Papers of the Year
MIT Robert M. Fano Award: Best EECS UROP Project
