Data centers are often large, complex systems, yet only a tiny fraction of the code running there handles the majority of requests. Thus, in order to make networking applications more efficient, optimizing the most trafficked lines of code is most beneficial. LakePlacid is a framework which has been proposed to serve this purpose. It succeeds in achieving microsecond-scale latency for data centers’ most frequent use cases, but the most frequent use case may shift over time. This year, I will research methods to continuously update kernel modules on the fly based on profile-guided optimizations.

I am excited to gain more research experience in computer systems and improve my technical communication. As someone who spent years coding without truly understanding the underlying systems, I find compilers fascinating, and I have enjoyed getting to learn more about them through classes and my past UROP. Through this SuperUROP, I am looking forward to learning more about networking and compilers through real-world research experience.