Hello there!

I'm a fifth-year PhD student at Northwestern University and was a visiting student researcher at Stanford University for the 2023-2024 academic year. I am advised by Professor Jason Hartline.

My primary research is in mechanism design, though my background as an undergraduate was in machine learning and economics. Somewhat idiosyncratically for someone in CS theory, I'm also knowledgeable about low-level/assembly programming.

Here are some projects I've worked on:

GPU Allocation

It takes compute resources to train machine learning models, such as LLMs. An enterprise that rents costly GPUs from a cloud provider needs some way to provision them to its employees on demand, who can then use them to train models.

On-demand provisioning should ideally

1. Leave few rented GPUs idle, since they have already been paid for. Unprovisioned GPUs represent wasted resources that the enterprise would have preferred not to rent.
2. Rarely prevent employees from accessing GPUs in periods of high demand. It's desirable to rent enough GPUs, or effectively ration the ones the enterprise has, to make supply constraints a non-issue.

There is a fundamental tradeoff between these two properties. A project with Adobe Research studying that tradeoff would eventually become the basis for Fundamental Limits of Throughput and Availability, published in EC 2024.

Untying Knots with Neural Networks

As an undergraduate, I did research with Professor David Bachman, a topologist at Pitzer College. We analyzed whether neural networks with custom-designed layers could correctly perturb heavily tangled polygonal knots until they became visually indistinguishable from the unknot.

This was an interdisciplinary project that combined insights from knot theory and machine learning. The final paper can be found here.

---

Other Projects

This is just a small sample of my broader research profile. I've invented a compression algorithm, investigated the Bayesian characteristics of prediction markets, and worked on various other interdisciplinary projects between economics and computer science.

I've acquired some amount of internet fame for my work in assembly programming and low-level optimization, too.

If you have questions about my work or are interested in collaborating, please reach out to me via email!