About Me

Hi! I am a Ph.D. candidate at the SNU Computer Vision Lab, advised by Prof. Bohyung Han.

I am interested in Video-Language Models, particularly how they represent and retain information over time. My previous work includes robust learning of temporal dynamics ^[SLT], efficient context modeling ^[TC-CLIP], and mechanistic interpretability of VideoLLMs ^{[Map the Flow]}. I am currently working on retaining long-term visual context under tight memory budgets.

I’m seeking Research Scientist opportunities starting in 2027.

CV · minji@snu.ac.kr

🤗 News

Jul 2026I joined LG AI Research as a Research Scientist Intern in the Physical Intelligence Lab.

Jan 2026Our paper, Map the Flow, has been accepted to ICLR 2026! This is one of the first attempts to understand the internal mechanisms of VideoLLMs through controlled experiments.

Oct 2025I was selected as a Top Reviewer in NeurIPS 2025.

Jul 2025I joined Amazon Berlin as an Applied Scientist Intern.

Older News

Feb 2025: I received the Best Paper Award (Grand Prize) at the Workshop on Image Processing and Image Understanding (IPIU) in Jeju, Korea.
Dec 2024: I have been selected as a winner of the Qualcomm Innovation Fellowship Korea 2024.
Jul 2024: Our paper on enhancing CLIP's video understanding capability by infusing global information has been accepted at ECCV 2024.
Oct 2023: I joined NAVER AI Lab as a research intern.
Jul 2022: Our paper on resolving training/testing inconsistency problem in visual tracking has been accepted at ECCV 2022.
Apr 2022: Our paper on online hybrid lightweight representation learning for visual tracking has been accepted at IJCAI 2022.

💼 Work Experience

LG AI Research (Seoul, Korea)

Jul 2026 — Present

Research Scientist Intern, Physical Intelligence Lab

Mentor: Pyunghwan Ahn

Amazon (Berlin, Germany)

Jun 2025 — Dec 2025

Applied Scientist Intern, Premium AI Science Team

Mentor: Matthieu Guillaumin

NAVER AI Lab (Seongnam, Korea)

Oct 2023 — Apr 2024

Research Intern, Backbone Research Team

Mentors: Taekyung Kim, Dongyoon Han

📄 Publications

Map the Flow: Revealing Hidden Pathways of Information in VideoLLMs
Minji Kim*, Taekyung Kim*, Bohyung Han (*Equal Contribution)
ICLR 2026 MapTheFlow_teaser

Leveraging Temporal Contextualization for Video Action Recognition
Minji Kim, Dongyoon Han, Taekyung Kim*, Bohyung Han* (*Corresponding Authors)
ECCV 2024 TC_CLIP_framework

Towards Sequence-Level Training for Visual Tracking
Minji Kim*, Seungkwan Lee*, Jungseul Ok, Bohyung Han, Minsu Cho (*Equal Contribution)
ECCV 2022 SLT_framework

Online Hybrid Lightweight Representations Learning: Its Application to Visual Tracking
Ilchae Jung, Minji Kim, Eunhyeok Park, Bohyung Han
IJCAI 2022 hybrid_framework

🏅 Honors and Awards

Top Reviewer, NeurIPS 2025
Qualcomm Innovation Fellowship Korea, 2024
IPIU Best Paper Award, 2023 & 2025
IPIU Undergraduate Best Paper Award, 2019
Intel FPGA Design Contest, 2017
Academic Excellence Scholarship, Konkuk University, 2016 - 2019

🎓 Education

Seoul National University
Ph.D. in Electrical and Computer Engineering (Advisor: Prof. Bohyung Han)
Mar 2020 — Feb 2027 (expected)

Konkuk University
B.S. in Electrical and Electronics Engineering
Mar 2016 — Feb 2020

🔍 Academic Service

Reviewer
Conference: CVPR, NeurIPS, ICLR, ICCV, ECCV, AAAI, WACV, BMVC
Journal: TPAMI, MVA

Teaching Assistant
Samsung AI Expert, 2024
Dependable Deep Neural Networks, Seoul National University, Spring 2022
Data Structures, Seoul National University, Fall 2021
Samsung AI Academy Advanced Course, 2020

🤝 Leadership

President of BOOT&4DIM
Electronics Engineering Academic Club, Konkuk University, 2018