Jongse Park

Jongse Park is an associate professor in the School of Computing (SoC) at KAIST and a faculty member of the Computer Architecture and Systems Laboratory (CASYS). He is co-affiliated with the Department of AI Computing, Department of Semiconductor System Engineering, and the Graduate School of AI Semiconductor. His research interests include Computer Architecture and Systems, Generative AI Serving Systems, On-Device Physical AI Systems, Processing-in-Memory (PIM) Acceleration, AI Compiler, Runtime, and Framework. He holds a PhD in Computer Science from Georgia Institute of Technology and is affiliated with the Pervasive Parallelism Lab at Stanford EECS as a Visiting Associate Professor.

Jongse Park is a researcher in Computer Architecture, focusing on HW/SW Codesign and AI Systems. His work explores methods to accelerate deep neural networks through bit-level dynamic composition, FPGA-based implementations, and limited-precision analog computation. Park also develops frameworks like Tabla and Dnnweaver to enhance statistical machine learning performance. His research includes heterogeneous acceleration for large language models, approximate hardware design, and secure GPU memory techniques. Additionally, he investigates distributed training, inference scheduling, and quality tradeoffs in approximate acceleration. His contributions span both hardware and software co-design for AI systems, emphasizing efficiency and scalability.