About

I am a Master's student in Data Science at Seoul National University, focusing on Knowledge Editing, Mixture-of-Experts architectures, AI Agents, and Reinforcement Learning. My research interests include knowledge editing, mixture-of-experts, agentic AI systems, and reinforcement learning for language models.

Education

Seoul National University

Sep. 2024 – Aug. 2026

Seoul, Korea

M.S. in Data Science (Advisor: Yohan Jo)

Overall GPA: 3.83 / 4.3 (95.3 / 100)

Research Focus:

  • Knowledge Editing
  • Mixture-of-Experts
  • AI Agents
  • Reinforcement Learning

Publications:

  • "Scenario-Based Risk Evaluation Framework for Financial Agent Systems", MS Thesis, SNU, 2026
  • "KL for a KL: On-Policy Distillation with Control Variate Baseline", ICML 2026 Workshop
  • "TriBench-Ko: Evaluating LLM Risks in Judicial Workflows", ICML 2026 Workshop
  • "Context-Robust Knowledge Editing for Language Models", ACL 2025 Findings

Korea University

Mar. 2018 – Aug. 2024

Seoul, Korea

B.S. in Industrial and Management Engineering

Major GPA: 4.16 / 4.5

Overall GPA: 3.87 / 4.5 (93.7 / 100)

Relevant Coursework:

  • AI and Intellectual Property (A+)
  • Data structures and algorithms (A+)
  • Information System Design (A+)
  • Object-oriented programming (A+)

Academic Leaves:

  • Mar. 2020 – Feb. 2022: Served in the Republic of Korea Army
  • Mar. 2023 – Aug. 2023: Studied in Naver Boostcamp AI Tech NLP (5th)

Research & Relevant Experience

Center for Trustworthy AI

Sep. 2026 – Present

Convergence Research Center, Seoul National University, Seoul, Korea

  • Served as the technical lead supporting an interdisciplinary research collaboration.
  • Built a real-time LLM interview web system covering live chat interviews.

VAIV Company

Jul. 2025 – Dec. 2025

Research Intern

  • Conducted research on Mixture-of-Experts architectures.
  • Developed methods for integrating domain-specific LLMs into a single model.
  • Optimized expert routing for domain-specific LLM integration.

H&NLP Lab, Seoul National University

Dec. 2023 – Jul. 2024

Research Intern (Advisor: Yohan Jo)

  • Conducted research on knowledge editing in large language models.

Naver BoostCamp AI Tech NLP (5th)

Mar. 2023 – Aug. 2023

Naver, Seongnam, Korea

  • Studied overall NLP knowledge and development coding skills.
  • Carried out several intensive NLP projects.

KUBIG (Data Science Academic Society)

Jul. 2022 – Jun. 2023

Korea University

  • Learned about basic of Deep Learning & NLP; studied Time Series & Multivariate Data Analysis.
  • Carried out several DL & NLP based projects.

Projects

Model-based CUDA Kernel Optimization

Sep. 2025 – Dec. 2025

Team Leader & Presenter

  • Utilized deep learning architectures to optimize tensor blocks and streams.
  • Benchmarked across TPM, MatMul, and Reduction kernels, achieving up to 1.48x speedup on memory-bound tasks.

NBA Prediction System with LLM Explainability

Mar. 2025 – Jun. 2025

LLM Engineer & Prompt Engineer

  • Developed an LLM-based explanation interface to translate model predictions into user-friendly insights.
  • Designed a multi-turn conversational system with Streamlit and an MCP server integrating prediction outputs with external knowledge.

NLP-based Patent Search Pipeline Construction

Sep. 2023 – Dec. 2023

Team Leader & Presenter

  • Constructed a double-layer multi-label encoder architecture and built an end-to-end pipeline.

MRC Project

Jun. 2023 – Jul. 2023

Core Developer

  • Developed the core codebase and implemented curriculum learning to optimize model training.
  • Integrated Haystack and Elasticsearch to build a robust retrieval system.
  • Established a systematic pipeline with DFD, Git Flow, and a Gantt chart.

Celebrity Voice Bot Project

Mar. 2023 – Jun. 2023

Team Leader & Presenter

  • Developed a Next Token Generation model to emulate specific linguistic patterns.
  • Trained Glow-TTS and HiFi-GAN models and developed a web interface.

Dating Course Recommendation Chatbot Project

Dec. 2022 – Feb. 2023

Team Leader & Presenter

  • Fine-tuned KoGPT and implemented a rule-based chatbot.

Publications

Scenario-Based Risk Evaluation Framework for Financial Agent Systems

MS Thesis, Seoul National University, 2026

Gyubin Choi, Yohan Jo.

Paper

KL for a KL: On-Policy Distillation with Control Variate Baseline

ICML 2026 Workshop

Minjae Oh, Sangjun Song, Gyubin Choi, Yunho Choi, Yohan Jo.

Paper

TriBench-Ko: Evaluating LLM Risks in Judicial Workflows

ICML 2026 Workshop

*Haesung Lee, *Gyubin Choi, Eun-Ju Lee, So-Min Lee, YouKang Ko, Dogyoon Lim, Sung-Kyoung Jang, Yohan Jo.

Paper

Context-Robust Knowledge Editing for Language Models

ACL 2025 Findings

*Haewon Park, *Gyubin Choi, Minjun Kim, Yohan Jo.

Paper

Awards and Honors

Dec. 2023

Second Prize

Population Forecasting Model Accuracy Comparative Analysis Competition, LH, Korea

Nov. 2023

Chair's Award

KOSIPER, Korea

Feb. 2023

Outstanding Paper Presentation Award

KOSIPER, Korea

Feb. 2023

Capstone Excellence Award

Korea University, Korea

Feb. 2023

Academic Honors

Korea University, Korea

Skills and Techniques

Programming

Advanced Python, C++ | Moderate CUDA, C, R, MySQL, PostgreSQL | Novice SAS

Frameworks

PyTorch, Hugging Face, scikit-learn, TensorFlow

Certificates

ADsP (2023), SQLD (2023)

Languages

English (OPIc Intermediate High, 2025)

Leadership & Activities

Planning Team Leader

KUBIG Data Science Society

Dec. 2022 – Jun. 2023

Coordinated multi-member collaborations, workshops, and conference presentations.

Student President

Dept. of Industrial Management Engineering, Korea University

Mar. 2019 – Feb. 2020

Class of 2018 Representative

Dept. of Industrial Management Engineering

Mar. 2018 – Feb. 2019

Research Interests

Knowledge Editing Mixture of Experts AI Agents Reinforcement Learning