Sowoong Kim
I’m a Ph.D. candidate in Computer Science and Engineering at Ulsan National Institute of Science and Technology (UNIST), advised by Professor Woongki Baek. My research focuses on systems and software for high-performance and efficient generative AI, including on-device inference, with additional interests in computer security. I received a B.S. in Computer Science and Engineering with a minor in Electrical Engineering from UNIST.
Contact
Office: Room 607-2, Bldg. 106, UNIST, Ulsan, Republic of KoreaNews
- Deco accepted to ICCD 2026
- Harmonia accepted to Euro-Par 2026
- Activation Sequence Caching accepted to PACT 2024
Honors & Awards
- Selected for a NAVER PhD Fellowship · KRW 5 million
Publications
-
“Deco: High-Throughput MoE Inference through Decoupled Batching and Cooperative Execution,”in the Proceedings of the 44th IEEE International Conference on Computer Design (ICCD), Nov. 2026 (to appear)
-
“Harmonia: QoS-Aware and High-Throughput Generative Inference with a Single GPU,”in the Proceedings of the 32nd International European Conference on Parallel and Distributed Computing (Euro-Par), Aug. 2026 (to appear)
-
“Activation Sequence Caching: High-Throughput and Memory-Efficient Generative Inference with a Single GPU,”
-
“MARF: A Memory-Aware CLFLUSH-Based Intra- and Inter-CPU Side-Channel Attack,”
-
“DPrime+DAbort: A High-Precision and Timer-Free Directory-Based Side-Channel Attack in Non-Inclusive Cache Hierarchies using Intel TSX,”