My research focuses on explainable AI and interpretability of deep neural models, with a focus on natural language processing tasks and pretrained models. I have worked on developing evaluation and modeling techniques for large pretrained langauge models to provide faithful, yet readable, explanations for their decisions. I am also interested in improving model robustness to out-of-distribution shifts and adversarial perturbations, again focused on natural language processing tasks and pretrained models. I have worked on generating data for counterfactual reasoning and identifying hidden dataset stratifications for distributional robustness.