Computer System Architecture Lab

Taewoon Kang

Ph.D. Student in Computer Science and Engineering

Ph.D. student at Korea University researching computer architecture, AI accelerators, and processing-in-memory systems. My work centers on efficient hardware and system design for modern machine learning workloads, with experience spanning PIM architecture, ASIC design, RTL implementation, and architectural simulation.

Get in touch View publications
Affiliation
Korea University
Based in
Seoul, Korea
Focus
Memory-centric systems

Research direction

Bringing computation closer to data.

I design memory-centric architectures that reduce data movement and make irregular, bandwidth-bound workloads run efficiently.

01

AI Accelerator

02

HW/SW Co-Design SoC

03

Processing for AI (NPU)

04

Computer Architecture

05

Processing-in-Memory (PIM)

Selected work

Publications

International conference and journal publications in PIM, HBM, and hardware acceleration.

01
ATC '26 Conference 2026 To appear

FLASH: Fault-Led, Attested, Secure Hybrid-Key Checkpoint for Confidential GPU

Jonghyun Lee, Taewoon Kang, Pratheek B, Yongqin Wang, Rachit Rajat, Mengyuan Li, Gunjae Koo, Daniel Wong, Murali Annavaram

Proceedings of the 2026 ACM Annual Technical Conference

02
CAL Journal 2026

SoftmaxPIM: An HBM-Based PIM Architecture for Accelerating GeMV-Softmax Execution Pipeline

Taewoon Kang, Namhun Kim, Jihun Lee, Murali Annavaram, Gunjae Koo

IEEE Computer Architecture Letters

DOI PDF
03
JSA Journal 2026

SparsePIM+: Accelerating SpMV on HBM-Based PIM via Logic-Die Accumulators with Opportunistic TSV Utilization

Taewoon Kang, Namhun Kim, Geonwoo Choi, Taeweon Suh, Gunjae Koo

Journal of Systems Architecture

DOI PDF
04
ICS '26 Conference 2026

SumcheckPIM: An Efficient HBM-Based PIM Architecture for Linear Complexity Zero Knowledge Proofs

Sunchae Kim, Taewoon Kang, Sangwon Shin, Taeweon Suh, Yibin Yang, Gunjae Koo

ICS '26: Proceedings of the 40th ACM International Conference on Supercomputing

DOI PDF Slides
05
ICS '25 Conference 2025

SparsePIM: An Efficient HBM-Based PIM Architecture for Sparse Matrix-Vector Multiplications

Taewoon Kang, Geonwoo Choi, Taeweon Suh, Gunjae Koo

ICS '25: Proceedings of the 39th ACM International Conference on Supercomputing

DOI PDF Slides
Domestic publications 4
2025

Performance Analysis of a Per-Row Activation Counting Mechanism in DRAM

Seongpil Yang, Taewoon Kang · The Institute of Electronics and Information Engineers 2025 Summer Conference

2022

A Design and Implementation of MP3 Player using ARM Cortex-M0 DesignStart

Sang-un Jo, Tae-woon Kang, Yongwoo Kim · The Korean Institute of Electrical Engineers Workshop, Yeosu, Korea

2022

A Design and Verification of ARM Cortex-M0 DesignStart MCU System Using Open Source IP

Sang-un Jo, Tae-Woon Kang, Yongwoo Kim · ICS 2022, Jeonju, Korea

2022

An Implementation of CNN-based Real-Time Face Mask Detector using Jetson Nano

Taewoon Kang, Yongwoo Kim · JKAIS, Vol. 23

Where I have worked

Research &
experience

Academic and industry experience across architecture research, applied AI, and teaching.

Jan. 2026 - Present

Visiting Scholar

Ming Hsieh Department of Electrical and Computer Engineering, University of Southern California

Conducting collaborative research under the supervision of Prof. Murali Annavaram.

United States

Jun. 2023 - Aug. 2023

Intern

SK Telecom

Developed spam detection methods using AI.

Korea

2023 - 2025

Teaching Assistant

Digital Logic Design · Spring 2023, Fall 2023

Basic of SW · Spring 2025

Korea University

Foundation

Background

Mar. 2023 - Present

Korea University

Ph.D. in Computer Science and Engineering

Advisor: Gunjae Koo · GPA 4.35 / 4.5 · Seoul, Korea

Mar. 2019 - Aug. 2022

Sangmyung University

B.S. in System Semiconductor Engineering

Advisor: Yongwoo Kim · GPA 4.28 / 4.5 · Cheonan, Korea

Thesis: An Implementation of CNN-based Real-Time Face Mask Detector using Jetson Nano

Earlier project

2022 · Main Researcher

CNN-based Real-Time Face Mask Detector

Implementation of a real-time face mask detection system using CNN on Jetson Nano. Advisor: Yongwoo Kim.

Technical toolkit

programming languages

C (including Atmel and Keil)C++Python (PyTorch)Verilog 2001SystemVerilog

eda tools

Synopsys EDA (Semi-custom ASIC design)Xilinx FPGAAltera FPGA

hardware skills

ARM Cortex-M0/M3 IP implementationRISC-V Simple Processor designASIC/Semi-custom chip design and verification

simulation tools

GEM5DRAMsim3

coursework

Parallel Processing, AI Accelerator ArchitectureAdvanced Computer Architecture, Storage System ArchitectureARM Cortex-M Series IP ImplementationElectrical Circuit, Microelectronic CircuitSemi-Custom ASIC Design with Synopsys Design Tool, VLSI DesignDigital Logic Circuit, Computer Architecture and Design, Design with Verilog, SoC DesignSemiconductor Physics, Semiconductor Fabrication ProcessCalculus, Engineering Mathematics

Awards & honors

May. 2022

The top prize of CISC Career day

Sangmyung University

Mar. 2022, Sep. 2021

Header(in one's dept.) Scholarship

Sangmyung University

Sep. 2021, Mar. 2021

"BONGSA" Scholarship (Leader of Department)

Sangmyung University

Mar. 2021

Work Scholarship

Sangmyung University

Mar. 2021, Sep. 2020

Secondary Header(in one's dept.) Scholarship

Sangmyung University