Home

I am a Lecturer in the Department of Computer Science and Engineering (CSE) at the Islamic University of Technology (IUT), where I also earned my B.Sc. with Honors in May 2023. I completed my undergraduate thesis being affiliated with the Systems and Software Lab (SSL) under the supervision of Mohsinul Kabir.

My research examines how large language models (LLMs) reason, create, and behave in social settings. Underlying this work is a broader aim: to bridge the gap between human and machine cognition by understanding the capabilities these systems acquire, the mechanisms that support them, and the limits of their resemblance to human thought.

One strand of my work concerns mathematical and scientific reasoning: how reliably LLMs solve problems, and how their performance shifts with language, context, and inference-time strategies. Its creative counterpart is narrative generation, where I am currently exploring nonlinear storytelling with LLMs. Together, these directions span two complementary sides of cognition: structured problem-solving and creative expression.

A second strand lies at the intersection of computational social science and responsible AI. Here I investigate how framing, language, and alignment shape models' political behavior and moral decision-making, with particular attention to ethics, bias, and fairness.

To understand what drives such behavior, I also look inside the models, using mechanistic interpretability and activation-space interventions to explain and steer it. Beyond individual models, I explore multi-agent systems and agentic AI, studying how coordinated reasoning, retrieval, and verification can tackle complex tasks.

Alongside these central directions, I pursue projects in Bangla NLP and computer vision, from language resources to visual understanding. My publications give a fuller account of these lines of inquiry.