Charlotte Ka Yee Yan
Hi! I'm Charlotte, a CS student finishing up my last quarter at Stanford, where I've spent the last few years chasing problems I find genuinely hard. Lately that's meant computer use agents and long-horizon memory systems.
I grew up in Hong Kong, went to boarding school in England, and somehow ended up in California. When I'm away from a screen, I'm usually deep in a good book, cooking something elaborate with my friends, or out on a run pretending I know what I'm doing. You can find out more about the projects I've worked on (at SAIL and in my spare time) below!
Here is some work I've done,
AutoLibra: Agent Metric Induction from Open-Ended Human Feedback
paperTurning open-ended human feedback into fine-grained evaluation metrics for language agents, enabling automated, concrete, and task-agnostic self-improvement. ICLR 2026.
Agent Unit Tests
codeSmall, repeatable checks on agent behaviour version-controlled tests for the messy edges of computer-use agents.
Parallel Population Based RL
paperTraining agents in parallel using an evolutionary PBT on top of PPO in the Madrona batch simulator.
Symphony
paperDynamically create sub-agents on the fly to solve multifaceted tasks. SOTA in complex task benchmarks.
Agent Library
blog postA framework for generating and orchestrating an ever-evolving library of stateful, specialized agents.
Some pics of my dog, Percy,
And some books I've been reading
Nudge
Thaler & Sunstein
The Dark Forest
Liu Cixin
The Verrine Orations
Cicero
The Brain That Changes Itself
Norman Doidge
Words of Radiance
Brandon Sanderson
Chip War
Chris Miller