I am a final-year Ph.D. candidate in Computer Science at the University of Virginia, advised by Prof. Felix Xiaozhu Lin. I build efficient and scalable infrastructure for AI workloads.
My research connects low-level systems design with the needs of modern AI applications. I work across operating systems and machine learning infrastructure to improve performance and resource efficiency, from datacenters to edge devices.
In my free time, I enjoy bouldering and taking photos with my old 35mm camera.
Ongoing projects
-
Fast data transfer in AI infrastructure (exploratory)
Exploring efficient data movement for reinforcement learning training, with potential directions including KV cache transfer and scheduling between rollout generation and training. -
Resource handover on Kubernetes (under submission)
Wonkyo Choe, Felix Xiaozhu Lin
Mitigating the performance degradation that AI workloads suffer during resource handover on Kubernetes.
Selected publications
-
SpMAP: Transparent Sparsity for LLMs (MobiSys’26)
Wonkyo Choe, Felix Xiaozhu Lin
[PDF] -
RWKV-Lite: Deeply Compressed RWKV for Resource-Constrained Devices (arxiv)
Wonkyo Choe, Yangfeng Ji, Felix Xiaozhu Lin
[PDF] -
AnA: An Attentive Autonomous Driving System (ASPLOS’25)
Wonkyo Choe, Rongxiang Wang, Felix Xiaozhu Lin
[PDF]
Other publications
-
Proto: A Guided Journey through Modern OS Construction (SOSP’25)
Wonkyo Choe*, Rongxiang Wang*, Afsara Benazir*, Felix Xiaozhu Lin
(* = equal contribution)
[PDF] -
Efficient NLP Inference at the Edge via Elastic Pipelining (ASPLOS’23)
Liwei Guo, Wonkyo Choe, Felix Xiaozhu Lin
[PDF] -
Rethinking Remote Memory Placement on Large-Memory Systems with Path Diversity (ApSys’21)
Wonkyo Choe*, Sang-Hoon Kim, Jeongseob Ahn
[PDF] -
Exploring the Design Space of Page Management for Multi-Tiered Memory Systems (ATC’21)
Jonghyeon Kim*, Wonkyo Choe, Jeongseob Ahn
[PDF] -
A Study of Memory Placement on Hardware-Assisted Tiered Memory Systems (CAL’20)
Wonkyo Choe*, Jonghyeon Kim, Jeongseob Ahn
[PDF]