I/O

in, out, and in between

Memos

coming soon — maybe

Monos

I occasionally write digestible, somewhat technical monographs on various subjects of interest, aiming to recapture the feeling of working through an especially illuminating textbook example or chapter. LLM coauthorship has made this easier!

Horizons

Intellectual themes I'd love to explore one day, usually motivated by papers I've recently read.

Dynamics of the acquired immune response

Our immune systems "learn" and "remember" in a very different way than our brains do. Can we write down the dynamic programming problem that our immune system solves to balance all of its objectives: resource efficiency, antigen specificity, antibody diversity, defense against future infections, etc.? Can we model and deconstruct each of the constituent "machines" of the immune system, such as the B-cell affinity maturation process, and then assemble each machine's contribution to each objective?

Inspiration: Merkenschlager, et al., Nature, 2025 and Pae, et al. Nature, 2025

Platonic convergence in representations of computation

There are many ways to represent information and perform computation. Do different information processing systems with similar selection pressures (e.g. resource constraints, market pressures, desired functionality) evolve isomorphic ways of representing information and implementing algorithms?

Inspiration: Evans, et al., Nature, 2024 and Huh, et al., arXiv, 2024

Unifying principles for the physics of life

Related to the above—instead of trying to build highly parametrized, noisy models of biology, can we instead isolate general principles at a higher level of abstraction that can be used to make quantitative predictions? In other words—can we write down "Newton's laws" for life?

Inspiration: Ravasio, et al., arXiv, 2024 and Bialek, arXiv, 2024

Deconstructing biological neural networks

Can we infer the effective weights and activation functions of biological neural networks by measuring intracellular, sub-threshold potentials and mapping them to standard electrophysiological measurements of those same neurons firing? Understanding biological networks this way would would then allow us to simulate them in silico. This was going to be my PhD project, the first result of which was published by the Park and Ham groups at Harvard in 2024.

Inspiration: Wang, et al., Nature Biomedical Engineering, 2024

Bookshelf

Now

reacquainting myself with Indian intellectualism

  • S. Radhakrishnan (Translations of the Upanishads and Dhammapada, Eastern Religion and Western Thought, Library of Living Philosophers)
  • J. Krishnamurti (The First and Last Freedom, Freedom from the Known)
  • Gitanjali, Rabindranath Tagore
  • The Satanic Verses, Salman Rushdie

Always

  • Finite and Infinite Games, James Carse
  • The Making of the Atom Bomb, Richard Rhodes
  • Gödel, Escher, Bach, Douglas Hofstadter
  • Mountains Beyond Mountains, Tracy Kidder: doing good need not be scalable to be right
  • Midnight's Children, Salman Rushdie
  • Blindness, José Saramago
  • Principles of Quantum Mechanics, Paul Dirac: it pissed me off that I graduated with an undergraduate physics degree not knowing what a photon was, so I chunked through this book during the first year of Sora. Now I do!

Someday

  • In Search of Lost Time, Proust: I would love to spend a year reading this with a few friends (text me if you would, too!)
  • Modern Classical Physics, Kip Thorne and Roger Blandford