Jennifer (Meng) Lu

I am an incoming PhD student in MIT EECS, where I will work with Prof. Marzyeh Ghassemi. I recently completed my Sc.M. in Computer Science at Brown University, working with Ellie Pavlick and Carsten Eickhoff

I am very interested in model interpretability and representation learning. My current research investigates the geometry of model representations, with a focus on understanding how geometric structure supports generalization, whether it can be used to predict model failures, and how these insights can inform more effective and robust training methods.

Outside of research, I enjoy creative writing, chamber music, tennis, reading, and learning about history (I majored in both CS and history during undergrad at Wellesley).