The fact that his research aligned so well with the work of his current advisor Chris Thomas, core faculty at the Sanghani Center, was a key influence in Zaber Ibn Abdul Hakim’s choice of where to pursue his master’s degree in computer science.

Hakim is working with multi-modal, multi-document information extraction and coreference resolution.

Additionally, he works with large language model (LLM) safety and adversarial attacks. In his research, Hakim has searched information across different modalities (text/video/audio) and across multiple documents in all possible combinations of query and search result modality.  

“Early in my research career, I understood the importance of joint modeling of multiple modalities instead of a single modality, as different modalities express complementary and important information about an event,” he said. “Continuing my research in this area, I have had the opportunity to explore it even further under Dr. Thomas. The practical implications of my current project made me more interested in this field of work.”

Hakim, who holds a bachelor of science degree in computer science and engineering from Bangladesh University of Engineering and Technology, said that being a graduate student at the Sanghani Center has provided him not only with relevant research, but support to pursue it.

“The center encourages research that is both technically rigorous and practically meaningful,” he said. “Additionally, I have benefited from the travel grant program that the center offers.”

Hakim’s published work includes:

  •  “LAMP: Learning Universal Adversarial Perturbations for Multi-Image Tasks via Pre-trained Models,” in main proceedings at the Association for the Advancement of Artificial Intelligence (AAAI) conference, 2026

He is projected to graduate in May of this year and is seeking an industry role in the artificial intelligence/machine learning field.