Liam Arzola

Computer Science PhD student at UC San Diego

Liam Arzola

I’m a PhD student in UC San Diego’s Systems and Networking group, advised by Amy Ousterhout and Alex Snoeren. I work on operating systems and datacenter infrastructure. My current research focuses on building software runtimes that schedule and offload microsecond-scale tasks, such as compression and signature verification, to on-chip accelerators.

In summer 2026, I interned with Microsoft Research and the Azure Systems Research Group. I benchmarked the overhead of snapshotting and restoring VMs when pausing coding-agent sandboxes between turns and identified optimizations to reduce snapshot latency by 88%. I implemented scatter-gather hypercalls and a decoupled VMM networking backend in Nanvix, and ran experiments on how state persistence affects coding-agent performance across task horizons.

Before joining UC San Diego, I was a software engineer in Microsoft’s Azure Core OS group, where I worked on VMBus and contributed to the OpenVMM and OpenHCL virtualization projects.

I previously worked with Antoine Kaufmann at the MPI-SWS Operating Systems Group on TCP stacks for virtualized environments and reconfigurable optical networks. I earned my B.A. in Computer Science at Cornell University, where I worked with Lorenzo Alvisi on Pesto, a Byzantine fault-tolerant SQL database.

selected publications

2026

  1. Tail Contagion: Sub-microsecond Time Protection in Shared Software Network Datapaths
    Matheus Stolet, Liam Arzola, Simon Peter, and Antoine Kaufmann
    arXiv preprint, 2026
    First posted September 2023. Revised April 2026.

2025

  1. Pesto: Cooking up High Performance BFT Queries
    Florian Suri-Payer, Neil Giridharan, Liam Arzola, Shir Cohen, Lorenzo Alvisi, and Natacha Crooks
    In Proceedings of the ACM SIGOPS 31st Symposium on Operating Systems Principles (SOSP), 2025