I am an assistant professor at the Department of Statistics & Actuarial Science at the University of Waterloo. I completed my PhD in Statistics at the University of Michigan working with Prof. Yuekai Sun and Prof. Moulinath Banerjee. Prior to my PhD, I completed my undergraduate and masters studies at Indian Statistical Institute, Kolkata with a focus on mathematics and statistics.
Research interests: I am broadly interested in the statistical foundations of data science. Some of my recent interests are:
- Distribution shift & transfer learning, Performative prediction
- Large Language Models: LLM routing, LLM alignment
- Fairness & Interpretability in ML algorithms
- Missing data
E-mail: smaity (at) uwaterloo (dot) ca
Recent News:
- Our recent work "Transfer Learning under Group-Label Shift: A Semiparametric Exponential Tilting Approach" has been posted on arxiv.
- Our recent work "Limitations of refinement methods for weak to strong generalization" on weak-to-strong generalization has been accepted in COLM 2025.
- Our recent work "CARROT: A Cost Aware Rate Optimal Router" has been accepted to ICLR 2025 FM-Wild Workshop.
- Our recent work "Estimation with missing not at random binary outcomes via exponential tilts." on non-ignorable missing data has been posted on arxiv.
- Our recent work "CARROT: A Cost Aware Rate Optimal Router." on LLM-routing has been posted on arxiv. Also, our router model and SPROUT benchmark dataset has been released in huggingface community.
- Our paper "Microfoundation inference for strategic prediction." [arxiv-preprint] has been accepted in AISTATS 2025.
- Our paper "Learning the distribution map in reverse causal in performative prediction." [arxiv-preprint] has been accepted in AISTATS 2025.
