I am a PhD student at the Halıcıoğlu Data Science Institute, UC San Diego.
From a causal perspective, my research centers on the principles and methods of modeling worlds, passively from the evidence they give, and modelers, which actively choose that evidence. My ultimate question is how to build a lifelong learning machine that continuously and efficiently improves and evolves in changing and novel worlds. My work spans a continual learning loop for causal world models: Explore (RSIAgent), Represent (TimeBraid), Identify (Transformer Is Inherently a Causal Learner, Causal-Copilot), and Generalize (WM3C).
Before joining UC San Diego, I worked with Dr. Konrad Kording on meta-learning methods for large-scale causal discovery in complex systems such as microprocessors. I am also interested in brain–computer interfaces and computational neuroscience, and previously worked on real-time neurofeedback systems with Dr. Gan Huang.
Understand, forecast, and shape time series with language.
A training-free multi-agent framework in which curriculum, actor and verifier agents explore a new digital environment, check what they learn, and keep it as reusable memory.
We reveal that transformers trained autoregressively naturally encode causal structures — gradient attributions directly recover underlying causal graphs without any explicit causal objectives.
An LLM-powered agent that runs the whole causal analysis loop — diagnosing the data, selecting and configuring the right method from 20+ options, checking its own results, and producing an inspectable report.
A reinforcement learning framework that generalizes to unseen environments by learning language-controlled causal components and recombining them — with identifiability guarantees.
Instead of designing a causal discovery algorithm, we learn one — from a microprocessor whose every causal edge can be established by intervention. It outperforms human-designed methods on silicon, simulated fMRI and gene networks.
A millisecond-level phase locked neural feedback system based on OpenBCI for real-time alpha wave regulation, integrating acquisition, phase estimation and stimulation on one chip.
Model
Paper
Paper
Video
Paper