Home Publications Projects Blog
|

Sejong Yang

Learner, Developer, and Researcher


2026.03
-
present
LG AI Research
I am currently working as an AI Scientist at LG AI Research (EXAONE Lab), contributing to EXAONE 4.5 — a hybrid VLM that unifies reasoning and non-reasoning capabilities in a single vision-language model — and applying it to vertical domain projects across finance, automotive, life science, and defense.
2019.09
-
2026.02
I received my Ph.D. in Computer Science from Yonsei University (Sep 2019 – Feb 2026), supervised by Prof. Seon Joo Kim. Thesis: Multi-Prior Alignment for Efficient Generative Vision Tasks. My research focused on face modeling with generative models (VAE, GAN, Normalizing Flows, Diffusion Models), 3D models (3DMM, Neural Renderer, NeRF), and multi-modal representation (visual-audio, visual-text).
2025.11
-
2026.02
LG AI Research
I spent the winter of 2025 as an AI Scientist Intern at LG AI Research (EXAONE Lab) in Seoul, just before completing my Ph.D., contributing to K-EXAONE — Korea's national sovereign AI foundation model — and EXAONE 4.5, a hybrid VLM that unifies reasoning and non-reasoning in a single vision-language model. Mentored by Soyeon Kim.
2023.05
-
2023.08
I was fortunate to be research scientist intern at Adobe Research in 2023 Summer at San Jose, California to solve the arbitrary talking head generation including human, animal, cartoon and pareidoliaface with StableDiffusion with my wonderful mentors Yang Zhou, Seoung Wug Oh, Cameron Smith and Deepali Aneja. Check projects tab for more information.
2020.06
-
2020.08
I have internship experience at NAVER in 2020 Summer to participate machine learning challenge named AI Rush.
2014.03
-
2019.08
I earned my BEng in Computer Science from Yonsei University. It was where I first learned what coding was, and where I completed my degree in the end. Outside of coursework, I was active in several clubs: served as Vice President of PoolC, the engineering college programming club; competed in poomsae events and performed with the Taekwondo demonstration team; and performed Croatian Rhapsody by Maksim Mrvica on a Steinway piano at Baekjunyeon Memorial Hall as a member of Piano In Yonsei.
2018.07
-
2019.08
I joined the startup Blammedia as CTO in 2019. We received a preliminary startup package investment worth KRW 30M from the Korea Institute of Startup and Entrepreneurship Development, and I learned the market value of technology and product through meetings with venture capitalists.
2018.01
-
2018.06
I spent a spring semester in 2018 as an exchange student at Uppsala University. Studying computer science in a different culture was a great experience.
2016.11
-
2017.11
In 2017, I worked at Ewoosoft as a developer on EzMobile and Ez3D-i. I learned how to use Xamarin.Forms, Qt5, and Flask for enterprise products, along with collaboration and DevOps tools such as Jira, Atlassian, Crucible and Fisheye, GitHub, and Jenkins in a waterfall development environment.
2016.08
-
2016.11
I have software assistant experience at Kakao in 2016 developing a component for openstack. In the experience, I learn the basics of the network, virtualization, and cloud computing from the experience.
2015.03
-
2019.08
As an undergraduate research intern in Prof. Seon Joo Kim's lab, I assisted with several color-related research projects: an auto white balance (AWB) display project, outdoor illumination observation with a hyperspectral camera, and data collection and refinement for color constancy tasks. I also worked on automatically finding baseball game highlights using video summarization methods.
Selected Publications and Projects
IF-MDM: Implicit Face Motion Diffusion Model for High-Fidelity Realtime Talking Head Generation
arXiv 2024.12
Sejong Yang, Yang Zhou, Seoung Wug Oh, and Seon Joo Kim.
[paper] [supplementary for videos]
Generating Paired Data beyond Human Face by Diffusion Feature-based Filter
2023.05 - 2023.08
Sejong Yang, Yang Zhou, Seoung Wug Oh, Cameron Smith, and Deepali Aneja.
[Technical Report] [Google Presentation]
THMM-DiT: Talking Head Motion Modeling with Diffusion Transformer
2022.11 - 2023.05
Sejong Yang, Seoung Wug Oh, Yang Zhou, and Seon Joo Kim.
[Technical Report] [Google Presentation]
Dense Interspecies Face Embedding
NeurIPS 2022
Sejong Yang, Subin Jeon, Seonghyeon Nam, and Seon Joo Kim.
[paper] [code] [project page]
Sheep Runner - Casual Unity Game
2020 Summer
Sejong Yang
Miscellanies