Projects
LLM-Assisted Explainability for Neural Combinatorial Optimization: Built an LLM-assisted explainability framework for neural combinatorial optimization using knowledge distillation and mixture-of-experts models, enforcing sparse decision policies for human-interpretable heuristics.
VAE-Based Causal Generative Model for Intracranial EEG: Implemented a VAE-based causal generative model to learn disentangled latent representations from intracranial EEG time-series, improving interpretability and downstream prediction accuracy.