Junwon Lee
Junwon Lee

Junwon Lee 이준원

M.S. Student @ IRIS Lab, Sungkyunkwan University, advised by Prof. Jong Hwan Ko 성균관대학교 IRIS Lab 석사과정 · 지도교수 고종환

Research interest: efficient AI inference — diffusion LLMs, KV-cache compression & quantization, hybrid attention, and hardware-software co-optimization. 연구 관심사: 효율적인 AI 추론 — diffusion LLM, KV-cache 압축·quantization, hybrid attention, 그리고 hardware-software co-optimization.

jwlee1218@g.skku.edu · GitHub · LinkedIn

News

  • Sep 2026 | Three papers accepted to the NeurIPS 2026 DiffuLM Workshop (1 orals, 2 poster)! NeurIPS 2026 DiffuLM Workshop에 논문 3편(Oral 1편, Poster 2편)이 채택되었습니다! Link
  • Mar 2026 | Starting my M.S. @ IRIS Lab, SKKU! 성균관대학교 IRIS Lab에서 석사과정을 시작합니다! Link
  • Feb 2026 | Graduated from Sungkyunkwan University — B.S. in Electronic and Electrical Engineering, magna cum laude. 성균관대학교 전자전기공학부 졸업 (magna cum laude).
  • Dec 2025 | RUSH accepted to RA-L 2026! RUSH가 RA-L 2026에 게재 승인되었습니다! Link

Publications

  • QWaF: Ultra-Low-Bit KV-Cache Quantization for Block-Diffusion Language Models

    Junwon Lee, Jae-sun Seo, Jong Hwan Ko · in collaboration with Cornell Tech

    NeurIPS 2026 Workshop on Diffusion Language Models (DiffuLM), Oral presentation

  • SPIRA: Sparse Pre-RoPE Informed Routing Attention for Diffusion Language Models

    Chaehyeon Min, Seokho Han, Junwon Lee, Jong Hwan Ko

    NeurIPS 2026 Workshop on Diffusion Language Models (DiffuLM)

  • TwinQuant: Decoding-Aware Dual-Centred Quantization for Diffusion Language Models

    Seokho Han, Chaehyeon Min, Junwon Lee, Jong Hwan Ko

    NeurIPS 2026 Workshop on Diffusion Language Models (DiffuLM)

  • RUSH: Recursive and Scalable 3D Coarse to Fine Path Planning

    Hwajung Lee, Daegeol Ko, Jaehyuk Hur, Junwon Lee, Seongbo Ha, Jong Hwan Ko, Hyeonwoo Yu

    IEEE Robotics and Automation Letters (RA-L), 2026

    Paper

Education

  • Sungkyunkwan University

    B.S.–M.S. Combined, Graduate School of Electrical and Computer Engineering

    Intelligent & Resource-efficient Image processing & Systems design (IRIS) Lab

    Advisor: Jong Hwan Ko

    2025 – present

  • Sungkyunkwan University

    B.S. in Electronic and Electrical Engineering (Magna Cum Laude)

    2020 – 2026

Experience

  • IRIS Lab, Sungkyunkwan University

    Research Intern — Multi-Modal LLM, Video Compression, Diffusion Quantization, Path Planning

    Advisor: Jong Hwan Ko

    Suwon, Korea | Jun 2024 – Sep 2025

  • DAVIAN Lab, KAIST

    Research Intern — 3D AI research

    Advisor: Jaegul Choo, Taewoong Kang

    Seongnam, Korea | Dec 2023 – Mar 2024

  • Whatcon

    Blockchain Engineer — smart contract & mainnet development

    Seoul, Korea | Sep 2023 – Jan 2024

  • Republic of Korea Army

    Sergeant, TOD Operator (military service)

    Sep 2021 – Mar 2023

Awards & Honors

  • First Prize - 2026 SKKU NPU Challenge

    Accelerated long-context decoding of Llama-3.2-3B on Tenstorrent Eagle-N NPU

    Sungkyunkwan University | Aug 2026

  • BK21 Scholarship

    Outstanding graduate students in ECE ($10k/year)

    BrainKorea21 | 2026-2027

  • Samsung Outstanding Poster Award

    Top-voted project, Industry–Academia Collaboration Exchange

    Samsung Electronics | Oct 2025

  • Best Project Award — SKKU Capstone Design Project

    FPGA-accelerated image super-resolution (FSRCNN) for hardware-software co-acceleration

    Sungkyunkwan University | Jun 2025

  • Dean's List — College of Information and Communication Engineering

    Sungkyunkwan University | Fall 2023, Fall 2024

  • Excellence Prize — 2023 Metaverse Developer Competition

    Personalized NFT metaverse gallery 'BID:AM'

    Ministry of Science and ICT, Republic of Korea | Oct 2023

  • Encouragement Prize — 2023 Blockchain Challenge for Solving Difficult Problems

    Ministry of Science and ICT, Republic of Korea | Aug 2023