Hojoon Leo Kim
MS Student, SNU ARC | Computer Science and Engineering, Seoul National University
hojoon.kim@snu.ac.kr
37, Samseong-ro 51-gil
Gangnam-gu, Seoul 06280
Republic of Korea
Hello, my name is Hojoon Kim, and I also go by Leo.
As a graduate researcher passionate about advancing ML systems through hardware-software co-design, my research focuses on agent-hardware co-design: jointly optimizing agentic algorithms for planning and memory with accelerator architectures and execution mechanisms such as plan reuse. I extend this full-stack approach to embodied AI and robotics, where on-device vision-language-action (VLA) models require real-time, energy-efficient inference. My work spans low-bit quantization, storage-assisted inference, and cache-driven planning for embodied AI agents, with publications at OSDI'25, ICML'25 (Spotlight), and MLSys'26. My long-term goal is to develop reusable system abstractions and open benchmarks that bridge ML algorithms and accelerator architecture, enabling scalable deployment of agentic and embodied AI.
When I’m not deep in code, you can probably find me on the tennis court🎾!
News
| Sep, 2026 | I’ve been selected for the YoulChon AI Young Researcher Fellowship, awarded by the Yulchon Foundation and Seoul National University AI Institute. |
|---|---|
| Jan, 2026 | Our work AgenticCache, a cache-driven asynchronous planning system for embodied AI agents, has been accepted to MLSys'26! Thank you Thierry and Yuheng! |
| Nov, 2025 | Our team won the Grand Prize (1st Place) at the 2025 AI Chip Contest - NPU Optimization Track hosted by the Ministry of Science and ICT (MSIT), Republic of Korea, receiving a prize of KRW 10,000,000. |
| May, 2025 | Our DecDEC has been accepted to OSDI'25! |
| May, 2025 | Excited to share that our ICML'25 submission, FlashTP, is currently under consideration for a Spotlight or Oral presentation. Also, it has already been successfully deployed in real-world industrial applications. |
Selected Publications
* indicates equal contribution
