I'm starting an M.S. at the KAIST Kim Jaechul Graduate School of AI in Fall 2026, in the MLAI Lab under Prof. Sung Ju Hwang.

Before that I built apps. I taught myself iOS during military service, led a small team to an App Store release, and wrote the backend for a campus platform with 500+ users. The problems I kept running into weren't ones I could fix by building another app, so I went into research instead. I work on coding agents now, mostly on whether the code they write is worth maintaining.

• Coding agents and multi-agent frameworks
• Masked diffusion language models
• Code maintainability

• Super Library Agent: Joint Generation and Maintenance of Multiple Applications Beyond the Single Codebase
  Daegyu Sung*, Yukyeong Lee*, Geon Park, Yumin Choi, Sung Ju Hwang
  Findings of EMNLP 2026
  A coding agent that writes a group of related applications over several rounds, pulls the code that repeats across them into one shared library, and rewrites every application written so far to import from it. On WebGen-Bench and PaperBench it cuts duplication, lines of code and description length without losing task accuracy.

• V-Agent: An Interactive Video Search System Using Vision-Language Models
  SunYoung Park*, Jong-Hyeon Lee*, Youngjune Kim*, Daegyu Sung, Younghyun Yu, Young-rok Cha, Jeongho Ju
  MMGenSR Workshop @ CIKM 2025
  A video search platform running three agents (routing / search / chat) over a fine-tuned VLM retrieval model, with zero-shot state of the art on MultiVENT 2.0. Done while interning at NC AI, where I worked on the agent pipeline and the Flutter client.

• porta — multi-agent investment portfolio feedback app, built on LangGraph
• Teambl — team-matching platform for KAIST; solo backend, iOS and Android, 500+ users
• 스르륵 — sleep management app; led a team of 3 from planning to App Store release

I also write at velog. Feel free to reach out. 🚀