# Peter Li

> Peter Li is a machine learning researcher at Northeastern University interested in statistical learning theory and large language model inference. His research aims to understand the mathematical and algorithmic processes of representation learning, compression, and memory in machine learning models. Contact: li.pet [at] northeastern [dot] edu.

Every page on this site has a plain-Markdown mirror at the same URL with a `.md` suffix.

## Profile

- [Peter Li - bio, education, teaching](https://peter.bio/about.md): Biography, education history, teaching record, and research interests.
- [Coursework at Northeastern](https://peter.bio/about/coursework.md): Complete graduate and undergraduate course list, 2020-2026.

## Research

- [Publications index](https://peter.bio/research.md): All publications.
- [InferScale: GPU-Native KV Injection for Personalized LLM Serving (2026)](https://arxiv.org/abs/2607.27090): preprint.

## Writing

- [Blog index](https://peter.bio/blog.md)
- [KV-Cache Reuse in Large Language Models (2026-05-20)](https://peter.bio/blog/kv-cache-reuse-in-large-language-models.md): An exploration on sparse attention and KV-Cache optimization for language models.
- [Hello World (2026-05-05)](https://peter.bio/blog/hello-world.md): Introducing my research blog.

## Optional

- [Sitemap](https://peter.bio/sitemap.xml)