Shuwen Sun

I am a Research Scientist working on AI infra at Meta. During my PhD, my research interests span broadly in the area of systems and networking, with a focus on building large scale distributed storage systems. Right now, I work on problems related to systems and infrastructure for AI.

I completed my Ph.D. from Khoury College at Northeastern in Oct 2025 and was advised by Peter Desnoyers. My dissertation, “Towards an Efficient and Strongly-consistent Distributed Object Store”, focused on the efficiency problem in distributed storage systems by introducing I/O amplification as a key metric to measure storage system efficiency (the ratio of total device I/O to client-requested I/O). In the dissertation, I then present ZStore, an efficiency-first distributed object store that achieves minimal I/O operations: one NVMe read per GET request and k writes per PUT request under k-way replication.

Prior to Northeastern, I earned my M.S. (2017) in Computer Science from Boston University under the supervision of Prof. Orran Krieger and was affiliated to Mass Open Cloud (MOC). During which my research revolved around performance diagnosis for distributed systems and datacenter networking. I also received a B.Eng. (2015) in Software Engineering from Sun Yat-sen University.