Undergraduate ResearcherSeoul, South Korea

Seungbeom Park

Machine learning research with an engineer's habits: measure, build, ship, repeat.

School of AI ComputingKAIST

Seungbeom Park
Seungbeom ParkUndergraduate Student, AI Computing

I am an undergraduate student in AI computing at KAIST. My work sits around machine learning: how models learn, where they fail, and how to turn what we learn about them into systems people can actually use. My first-author paper SkillCombiner was accepted to the REALM Workshop at EMNLP 2026, and ScatterPIN, a co-authored paper, was accepted to IEEE S&P 2027.

Outside of research I write software for its own sake, this site included. If any of this overlaps with what you are working on, my inbox is open.

Research interests

  • Machine Learning
  • Natural Language Processing
  • ML Systems
  • Human-AI Interaction

Signals

News

  • ScatterPIN, a co-authored paper on RF backscatter side-channel attacks, was accepted to IEEE S&P 2027.new

  • SkillCombiner was accepted to the REALM Workshop at EMNLP 2026 as a poster.

  • This site went live.

Research

Publications

2 total1 first author0 under review
AcceptedWORKSHOPFirst author

SkillCombiner: Optimizing Agent Skills via Edit Combination Search

Poster

Seungbeom Park, Jaewon Chu, Hyunwoo J. Kim

Second Workshop for Research on Agent Language Models (REALM @ EMNLP) · 2026

  • PDFsoon
  • arXivsoon
  • Codesoon
AcceptedCONFERENCE

ScatterPIN: PIN Inference Attack on Touch Keypad Door Locks via RF Backscatter Side Channel

Changgeon Kang, Dongjin Seo, Jonghyuk Yun, Seungbeom Park, Junyong Kim, Jun Han

IEEE Symposium on Security and Privacy (S&P) · 2027

  • PDFsoon
  • arXivsoon
  • Codesoon