Sanjana Ramprasad

I am a Ph.D. candidate in NLP/ML at Northeastern University, where I am fortunate to be advised by Byron C. Wallace.

My dissertation research focuses on evaluating factuality in large language models with a special focus on consequential domains.

I was lucky to spend multiple summers as a Research Intern at Abridge AI, working with Zachary Lipton and Elisa Ferracane on factuality in clinical dialogue summarization.

My interests have expanded to examining whether language models can reason faithfully under uncertainty. I study both epistemic uncertainty, which arises from gaps in knowledge, and aleatoric uncertainty, which emerges in scientific contexts where models must reconcile conflicting evidence and synthesize findings

I am also exploring confabulation control through mechanistic-interpretability methods, with the goal of proactively identifying and mitigating severe confabulations before they appear in generation.

Prior to my Ph.D., I also worked as a Senior Machine Learning Engineer at Mya Systems (2017-2020), where I contributed to the NLU components of the conversational AI system.