Luciano Maldonado

I’m a senior Computer Science student at West Virginia University, where I work in the Machine Intelligence Lab with Dr. Gyawali. My research has focused on the reliability and internal behavior of language models, including sycophancy, multimodal reasoning, and mechanistic interpretability. I’m interested in understanding how internal representations and computations relate to the predictions, reasoning, and behaviors models produce, and in developing methods that make those relationships easier to study.

I want to understand how post-training, especially reinforcement learning, reshapes the internal mechanisms that support reasoning, planning, and decision-making in autonomous agents. I’m particularly interested in how agents form and use representations of goals, actions, and their environments, how those representations influence the decisions they make, and how these questions extend to world models, where agents learn and plan through dynamical representations of their environments.