Hwijoon Lim

Hwijoon Lim (임휘준)

Research Scientist @ Meta

About Me

I am a Research Scientist at Meta, currently part of the AI & Systems Co-Design Team. I work on performance modeling and projection for GenAI training across Meta's hyperscale fleet, identifying bottlenecks and evaluating scaling strategies on GPUs and MTIA.

I received my Ph.D. degree from KAIST, where I was advised by Prof. Dongsu Han. During my Ph.D., I worked on improving machine learning and networking systems through designing efficient algorithms and implementing system-level optimizations in real-world systems. My contributions have been recognized at numerous top-tier venues, including SIGCOMM, ICML, and EuroSys. I received my B.S. degree in Electrical Engineering from KAIST, Republic of Korea.

Resume

Work Experience

Education

Publications

SAND: A New Programming Abstraction for Video-based Deep Learning

Juncheol Ye, Seungkook Lee, Hwijoon Lim, JiHyuk Lee, Uitaek Hong, Youngjin Kwon, Dongsu Han

ACM SOSP 2025 [Paper]

Efficient Disaggregated Cloud Storage for Cold Videos with Neural Enhancement

Jinyeong Lim, Juncheol Ye, Jaehong Kim, Hwijoon Lim, Hyunho Yeo, Junhyeok Jang, Myoungsoo Jung, Dongsu Han

IEEE Micro [Paper]

Awards

My Skills

C/C++, Python95%
C/C++, Python
Performance Modeling & Analysis90%
Performance Modeling & Analysis
ML Systems & Model Co-design85%
ML Systems & Model Co-design
Distributed Systems85%
Distributed Systems
Transport Protocols (TCP, RDMA)80%
Transport Protocols (TCP, RDMA)