Keynote Speaker
Prof. Donghyun Kang
Department of Computer Science and Artificial Intelligence at Dongguk University in Seoul, South Korea
Prof. Donghyun Kang is an Associate Professor in the Department of Computer Science and Artificial Intelligence at Dongguk University in Seoul, South Korea. His research lies at the intersection of system software, emerging storage technologies, and advanced computing infrastructures. Prof. Donghyun Kang’s pioneering work focuses on operating systems (OS), file and storage systems, NAND flash memory, Non-Volatile RAM (NVRAM), and system-level optimizations for Large Language Models (LLMs).
Throughout his academic career, he has made significant contributions to the development of next-generation storage systems, including Zoned Namespaces (ZNS) and innovative programming schemes designed to mitigate I/O interference. His research group actively pursues hardware-software co-design methodologies to build high-performance, low-latency, and energy-efficient systems for modern, data-intensive workloads.
Prof. Donghyun Kang received his Ph.D. from Sungkyunkwan University and collaborates closely with global semiconductor and cloud storage leaders to bridge the gap between academic theory and industry implementation.
Title:
Insights Gained from Next-generation Storage Solutions in AI Computing
The explosive growth of Artificial Intelligence (AI) and Large Language Models (LLMs) demands a fundamental shift in data architecture. To fully realize the potential of high-performance AI computing, modern infrastructure requires more than just raw computational power. This keynote addresses the critical storage bottlenecks faced in AI computing environments and shares key insights gained from overcoming these challenges. We will explore the architecture of ultra-fast data pipelines designed to eliminate I/O waiting times, strategies for building cost-effective yet scalable data lakes, and methods for ensuring predictable scalability. Finally, the session will provide a forward-looking roadmap on how hardware and software must converge to power the next generation of AI enterprises.