Sabyasachi Basu (সব্যসাচী বসু) 

PhD Student, Computer Science, 

University of California, Santa Cruz (UCSC)

E2-489, Baskin School of Engineering,

1156 High Street, Santa Cruz, CA 95064

Email: sbasu3 at ucsc dot edu 

I am a fifth year PhD in computer science at UC Santa Cruz (UCSC). My advisor is Prof. C. Seshadhri. Before my time at UCSC, I was an undergraduate majoring in mathematics at the Indian Institute of Science, Bengaluru. My advisors there were Prof. Vishwesha Guttal and Prof. Srikanth Iyer. My BS thesis was on developing a model for herding to explain catastrophic transitions in financial markets. A copy of my academic CV can be found here (last updated Feb 2021; there is a newer version of this available upon request, but you can piece together the new information by going through this website).

Because this comes up often, I prefer being called Sabya. For a long time most of my friends and colleagues called me by various versions of my last name, but I insist you call me Sabya (or, if you're familiar with Bangla, Shobbo). 

In the summer of 2023 was a student researcher at Google working with Aneesh Sharma to understand retrieval in graph embeddings and designing better training/retrieval procedures to obtain better results.

Current Research:

Broadly, my interests hover around discrete mathematics, probability, and their intersection. A lot of this ends up being graph theory, especially spectral graph theory, and most of these topics are of interest to people in CS/data science. I also enjoy studying dynamical systems, evolution, and logic. A fair bit of my work directly applies to use cases in recommendations and personalizations, both of which I have worked in (in different capacities).

Most of my recent research lies in a combination of designing efficient algorithms to decompose large graphs meaningfully, and understanding what structural/spectral properties of the graphs allow such decompositions to happen. I also care about making graph representations useful for different tasks, and identifying tradeoffs in ease of representation and utility for downstream objectives.