Along with my advisor, I work on problems that lie at the intersection of natural language processing (NLP) and deep learning. My research interests lie broadly in using unsupervised learning, semi-supervised learning, and transfer learning methods that exploit large volumes of unlabeled data or beneficial relationships among tasks to improve learning in target tasks with limited or no labeled data available. Research directions in this vein that I am currently exploring include: understanding and improving representations from pre-trained language models improving pre-training and fine-tuning methods (learning objectives, model architectures, training data, etc.) to make language models more robust and stable, especially in low-data scenarios exploring and improving transferability between NLP tasks leveraging data augmentation and self-supervised/semi-supervised learning methods to improve generalization and robustness exploring parameter-efficient approaches (e.g., prompt-based learning) for applying large-scale pre-trained language models to downstream tasks