Biography

I am a Ph.D. student at University of Pennsylvania, where I am fortunate to be advised by Prof. René Vidal. My PhD research centers on developing principled algorithms to control deep generative models at inference time. Towards this goal, I have worked on the following areas:

1. Language Model Steering: developing efficient guidance algorithms to steer diffusion LLMs using downstream reward models, controlling LLMs via activation steering and concept decompositions.
2. Diffusion Model Steering: using uncertainty estimates to improve posterior sampling for image diffusion models (pixel and latent space), with applications in image restoration and medical imaging.
3. Safety and Robustness: optimization-based algorithms for eliciting hallucinations and jailbreak attacks for LLMs, developing defenses for adversarial attacks in computer vision.

I was an AWS-AI ASSET Fellow at UPenn for the year of 2024. I previously completed my MS in Computer Science at Columbia University, where I was advised by Prof. John Wright, and my BS degree in CS and Math at UT Austin.

I am actively looking for full-time Research Scientist positions starting in early 2027.