Prachi Goyal

I’m a graduate student in the Language Technologies Institute at Carnegie Mellon University, where I’m exploring the frontiers of natural language processing and machine learning systems. Before joining CMU, I spent time at Microsoft building scalable backend systems as a software engineer. I completed my undergraduate studies at IIIT Delhi in 2023, earning a degree in Computer Science and Applied Mathematics.

My research centers on machine learning and natural language processing, particularly in areas of explainability and optimization. I’m drawn to questions around efficient inference, interpretability in sequence models, reinforcement learning, and foundation models—essentially, understanding how to make powerful AI systems both effective and understandable.

During my time at IIIT Delhi, I had the opportunity to work with Prof. Ranjeetha Prasad on federated learning research. Together, we developed an algorithm that enables federated learning systems to better utilize unannotated data, which was published at ICASSP 2024. Currently at CMU, I’m collaborating with Prof. Gauri Joshi on optimizing inference for large language models. Our work focuses on developing techniques to make LLM inference more efficient without sacrificing performance, addressing the growing computational demands of deploying these models at scale.

When I’m not immersed in research, you’ll likely find me on a court or field—I’m an avid player of badminton, table tennis, and lawn tennis. I also enjoy singing and occasionally pick up a pencil to sketch.