Hi!

I am a first-year PhD student in Computer Science at Columbia University, advised by Dave Blei, Hongseok Namkoong, and Richard Zemel.

I am interested in ML and AI, with a focus on understanding and improving out of distribution generalization and robustness under distribution shift across domains, including language. Right now my main interests are

Trustworthy AI agents
Making agents reliable on long horizon tasks by improving how they meta-learn and giving them memory models that let them learn from their own experience.
Continual learning
Building models that keep learning from new data and shifting distributions over time, without forgetting what came before.

If this overlaps with what you are working on or thinking about, feel free to send me an email to chat or explore possible collaborations.

Publications

2026

Mittal D., Castellani T., Yen T., Ye N., Wu F., Chen M., Cai T., Koukoumidis E., Zeng W., Namkoong H. LatentGym: A Testbed for Cross-Task Experiential Learning with Controllable Latent Structure. Under review.

2025

Castellani T., Ye N., Mittal D., Yen T., Koukoumidis E., Zeng W., Namkoong H. SynthTools: A Framework for Scaling Synthetic Tools for Agent Development. Under review.