Portrait
Junggeun Do
MSCS Student
Texas A&M University
About Me

I'm a Master's student in Computer Science at Texas A&M University, advised by Prof. Kuan-Hao Huang. I previously earned my BS in Physics and BA in Economics from Seoul National University, along with a minor in Computer Science.

My current research focuses on multilingual large language models, particularly on understanding how these models represent and process knowledge across languages. I am especially interested in language alignment and interpretability, studying how these models encode language-specific and shared knowledge and how these internal mechanisms influence behavior. Recently, I have been working on model editing and machine unlearning in multilingual settings.

Before beginning my graduate studies, I worked as a Machine Learning Engineer across diverse environments, from early-stage startups to large enterprises in the Korean tech industry. During this time, I improved machine translation models to better match domain-specific data distributions, developed personalized TTS models using limited data, and built real-time search systems for e-commerce platforms. These experiences shaped my interest in bridging machine learning research with real-world applications and building reliable deep learning systems.

Education
  • Texas A&M University
    Texas A&M University
    M.S. in Computer Science
    Aug. 2025 - Present
  • Seoul National University
    Seoul National University
    B.S. in Physics ยท B.A. in Economics
    Minor in Computer Science
    Aug. 2024
Experience
  • Blux
    Blux
    Machine Learning Engineer
    Jan. 2025 - Jul. 2025
  • Seoul National University
    Seoul National University
    Research Intern
    Apr. 2023 - Jan. 2025
  • Gravity Labs Co., Ltd.
    Gravity Labs Co., Ltd.
    Software Engineer
    Sep. 2022 - Mar. 2023
Publications
SEAL-pose: Enhancing 3D Human Pose Estimation via a Learned Loss for Structural Consistency
SEAL-pose: Enhancing 3D Human Pose Estimation via a Learned Loss for Structural Consistency

Yeonsung Kim*, Junggeun Do*, Seunguk Do, Sangmin Kim, Jaesik Park, Jay-Yoon Lee (* equal contribution)

arXiv preprint (earlier version presented at ICCV 2025 Workshop SP4V)

SEAL-pose: Enhancing 3D Human Pose Estimation via a Learned Loss for Structural Consistency

Yeonsung Kim*, Junggeun Do*, Seunguk Do, Sangmin Kim, Jaesik Park, Jay-Yoon Lee (* equal contribution)

arXiv preprint (earlier version presented at ICCV 2025 Workshop SP4V)

ContrastiveMix: Overcoming Code-Mixing Dilemma in Cross-Lingual Transfer for Information Retrieval
ContrastiveMix: Overcoming Code-Mixing Dilemma in Cross-Lingual Transfer for Information Retrieval

Junggeun Do, Jaeseong Lee, Seung-won Hwang

NAACL 2024 Oral

ContrastiveMix: Overcoming Code-Mixing Dilemma in Cross-Lingual Transfer for Information Retrieval

Junggeun Do, Jaeseong Lee, Seung-won Hwang

NAACL 2024 Oral

All publications