Spotting Problems for Auto Research
How to recognize the parts of research that become combinatorial search—and where parallel research agents can make real progress.
I’m a PhD student in computer science at Princeton, advised by Sanjeev Arora.
My research focuses on how machine learning systems generalize beyond their training data. I’m particularly interested in self-improving models, synthetic data, and automated research.
Latest
Writing
Stuck Agents: Optimization Plateaus and the Case for Context Resets
An empirical study of long-horizon coding agents, optimization plateaus, and context-reset interventions.
Length and Composition Generalization: Sample Complexity
A research note exploring the sample complexity of length and composition generalization in machine learning models.
Research
I combine controlled synthetic problems, empirical scaling, and theory to study what models learn and what they can discover next.
Generalization beyond the training distribution
Understanding how learning systems extrapolate across length, hardness, and composition—and what their successes reveal about learned algorithms.
Self-improving learning systems
Studying synthetic data, weak-to-strong scaling, and feedback loops that let models create increasingly useful supervision.
Automating the research loop
Finding research problems where hypothesis generation becomes structured search, then designing agents that can explore that space in parallel.
Selected papers
Full CV →Extrapolation by Association: Length Generalization Transfer in Transformers
Ziyang Cai, Nayoung Lee, Avi Schwarzschild, Samet Oymak, Dimitris Papailiopoulos
Self-Improving Models Overcome Length and Hardness Generalization via Weak-to-Strong Scaling
Ziyang Cai*, Nayoung Lee*, Avi Schwarzschild, Kangwook Lee, Dimitris Papailiopoulos
Everything Everywhere All at Once: LLMs Can In-Context Learn Multiple Tasks in Superposition
Zheyang Xiong, Ziyang Cai, John Cooper, et al.
Delving into Out-of-Distribution Detection with Vision-Language Representations
Yifei Ming, Ziyang Cai, Jiuxiang Gu, Yiyou Sun, Wei Li, Yixuan Li
Background
Before Princeton, I studied at UW–Madison and built automated synthetic-data pipelines for coding agents at Microsoft Research. I was previously advised by Dimitris Papailiopoulos, Junjie Hu, and Sharon Li.