hs-yang.com

Hyunseo Yang

Hyunseo Yang

I'm a B.S. candidate in Computer Science & Engineering and Business Administration at Seoul National University.

My research interest lies in artificial intelligence and large language models, together with human-computer interaction — how intelligent systems can support the way people create, decide, and learn.

I'm currently a research intern at the Algorithmic Intelligence Lab @KAIST, advised by Prof. Jinwoo Shin. Previously, I was a research intern at HCIL @SNU, advised by Prof. Jinwook Seo, and interned at DeepAuto.ai.

News

Started a research internship at ALIN Lab @ KAIST

Jul 2026

Working on world-model supervision for computer-use agents, advised by Prof. Jinwoo Shin.

DirectVis accepted to IEEE PacificVis 2026!

Jan 2026

Our short paper on editing code-based visualizations with direct manipulation.

Second place at the Future Finance Talent Competition

2026

Blockchain division, hosted by Hanwha Life.

Started a research internship at HCIL @ SNU

Dec 2025

Advised by Prof. Jinwook Seo at the Human-Computer Interaction Lab.

First place at the Deep Learning Hardware Design Contest

2025

POLARIS Next-Generation Semiconductor Convergence Innovation University Program.

Publications

2026

Scaling World-Model Supervision for Computer-Use Agents

Hyunseo Yang, Byungwoo Jeon, Jinwoo Shin

Working Paper

DirectVis: Editing Code-Based Interactive Visualization with Direct Manipulation

Jeongin Park, Mingyu An, Hyunseo Yang, Junhyeong Hwangbo, Min Hyeong Kim, Hyeon Jeon, Jinwook Seo

IEEE PacificVis 2026 · Short Paper

Experience

Algorithmic Intelligence Laboratory, KAIST

Research Intern · Seoul, Korea

Jul 2026 – Present

  • First-authoring Scaling World-Model Supervision for Computer-Use Agents.
  • Leading the data-scaling subproject for the LLM-based world model of a computer-use (GUI) agent, covering supervision synthesis, curation, and scaling.

Human-Computer Interaction Lab, Seoul National University

Research Intern · Seoul, Korea

Dec 2025 – Jun 2026

  • Co-authored DirectVis, accepted as a short paper at IEEE PacificVis 2026.
  • Co-authoring "Revisiting Map Projections for Geospatial Data Analysis: Task Taxonomy, Empirical Study, and Guideline" for submission to IEEE TVCG 2027.
  • Led the development of an interactive diffusion-based image editing system that supports user intervention during denoising.

DeepAuto.ai

Intern · Seoul, Korea

Jan 2025 – Feb 2025

  • Led the development of the DeepTutor.ai demo platform showcasing long-context reasoning technology for clients and investors.
  • Built the end-to-end front-end and performed prompt engineering to improve system robustness and interaction reliability.