Karan Srivastava

Karan Srivastava

PhD Student, Mathematics

University of Wisconsin-Madison


I’m a fourth year PhD student in mathematics advised by Jordan Ellenberg. I’m interested in how we can apply Reinforcement Learning (RL) - a field of machine learning where we train a computer agent to learn how to take actions in an environment - to problems in pure mathematics and science to help mathematicians and scientists conjecture, discover, and prove new truths in their fields.

Currently, my research focusses on creating generative RL to generate useful, interpretable data in additive combinatorics for mathematicians in this area to use to inform conjectures and proofs. I’m excited by what mathematics and science will look like in 100 years and the ai-powered tools academics in the future will have access to, and I’m driven to do my small part in this rich field of possibilites. I’m also affiliated with the Wisconsin Institute for Discovery and the Institute for Foundations of Data-Science.

I was born in Kolkata, India - a city rich in history and culture with bustling streets and great food - and lived there until I moved to the calmer lands of Illinois, where I got my BSc in Mathematics at the University of Illinois at Urbana-Champaign. I also spent a semester at the Independent University of Moscow as a part of the Math in Moscow study abroad program.

Download my resumé.

  • Machine Learning
  • Reinforcement Learning
  • Combinatorics
  • Algebraic Geometry
  • PhD in Mathematics, 2025

    University of Wisconsin-Madison

  • BSc in Mathematics, 2020

    University of Illinois at Urbana-Champaign


PhD in Mathematics
University of Wisconsin-Madison
Aug 2020 – Present Madison
I am currently interested in how we can apply techniques in reinforcement learning to help pure mathematicians. Given a quesiton in mathematics, can we get a machine to learn how to generate useful, interpretable, and generalizable computations in that space that can give us some insight about how we can tackle the problem. I’ve also looked at how we can reconstruct linear subspaces using techniques in algebraic geometry. I am doing a graduate minor in computer science.
BSc in Mathematics
University of Illinois at Urbana-Champaign
Aug 2016 – May 2020 Urbana-Champaign
I took various graduate level courses in algebraic geometry, commutative algebra, and algebraic topology at UIUC. I had the opportunity to work on research projects with Susan Tolman, studying Cremona transformations on Symplectic Manifolds, Dominic Culver, in studying Weierstrass equations for elliptic curves, and Zoi Rapti, in studying mathematical models of coevolutionary biology.
I also took some time to explore the math world in a literal sense and participated in the Math in Moscow study abroad program in my third year - a challenging and engaging experience that shaped my early interest in algebraic geometry.


(2022). A Perturbation Bound on the Subspace Estimator from Canonical Projections. IEEE ISIT 2022.



Reinforcement Learning for Generating Useful Combinatorial Data
Optimal Subspace Estimation with Noise
A Perturbation Bound on the Subspace Estimator from Canonical Projections
An 'almost impossible' puzzle and group theory
Why people say "I can't do math"


Here are some organizations I’m … organizing.

Madison Experimental Mathematics Lab
UW - Madison, Madison, WI
Undergraduate Mentor Program
UW - Madison, Madison, WI
Directed Reading Program
UW - Madison, Madison, WI
Math Circle
UW - Madison, Madison, WI


I love teaching! Here are the classes that I’ve taught as a teaching assistant:

Math 221 - Calculus and Analytic Geometry 1 - Fall 2020
Math 211 - Calculus - Spring 2021, 2022, Fall 2021, 2022
Math 222 - Calculus and Analytic Geometry 2 - Summer 2021
Math 240 - Intro to Discrete Mathematics - Summer 2022, 2023

Here are worksheets that I’ve written.