Hanyong Lee

AI Researcher / Ph.D. Student

Toggle navigation menu

Researcher profile / trustworthy AI and research systems

Hanyong Lee이한용

Building trustworthy NLP and autonomous research systems

I build language systems that remain useful under noisy, adversarial, and real-world conditions, and I care as much about disciplined execution as I do about model quality.

PublicationNAACL 2025BitAbuse dataset for visually perturbed phishing text defenseSystemAutoLabOSGoverned operating loop for autonomous research execution
FocusTrustworthy NLPRobust language systems under noisy and adversarial conditions

Research focus

  • Trustworthy NLP
  • Autonomous Research Systems
  • LLM-based Applications
  • Experiment Governance
  • Robust Language Understanding

Mission

My work connects trustworthy NLP, LLM applications, and research infrastructure, with an emphasis on governed workflows, evidence-aware evaluation, and AI systems that can be inspected, resumed, and trusted.

RolePh.D. Student in Artificial Intelligence

Chung-Ang University

LabCAU AutoML Lab

Seoul, Korea

Emailglhy0718@gmail.comORCID0009-0001-6900-4638Google ScholarHanyong LeeBloglhy0718.github.ioHugging FacelhyX@hanyong_lee_

Research focus

  • Trustworthy NLP
  • Autonomous Research Systems
  • LLM-based Applications
  • Experiment Governance
  • Robust Language Understanding

Mission

My work connects trustworthy NLP, LLM applications, and research infrastructure, with an emphasis on governed workflows, evidence-aware evaluation, and AI systems that can be inspected, resumed, and trusted.

Overview

Research

The thread running through my work: robust language understanding, governed experiments, and systems that make research execution inspectable.

Current focus

Trustworthy NLP, autonomous research execution, and evidence-aware LLM systems

Research style

Turning research ideas into governed, inspectable systems instead of one-off demos

What I build

Research infrastructure, evaluation workflows, and real-world NLP applications

Based in

Seoul, Korea

Selected publications

Representative publications

Representative publications across phishing defense, adversarial robustness, dialogue systems, and practical language model development.

Neurocomputing

2026

Survey on Joint Compression and Fine-Tuning of Large Language Models: Methods, Toolchains, and Open Challenges under Resource Constraints

J. Kim, H. Lee, J. Lee

Article 134332 | doi: 10.1016/j.neucom.2026.134332

View publication

Findings of NAACL 2025

2025

BitAbuse: A Dataset of Visually Perturbed Texts for Defending Phishing Attacks

H. Lee, C. Lee, Y. Lee, J. Lee

New Mexico, USA | April 29 - May 4, 2025

View publication

Computers, Materials & Continua

2025

OCR-Assisted Masked BERT for Homoglyph Restoration towards Multiple Phishing Text Downstream Tasks

H. Lee, Y.-C. Park, J. Lee

Vol. 85, No. 3, pp. 4977-4993 | doi: 10.32604/cmc.2025.068156

View publication
View 4 additional publications

ICCE 2025

2025

GoodGPT: Counseling-Chat

K. Kim, H. Lee, J. Lee

Las Vegas, USA | January 11 - 14, 2025 | doi: 10.1109/ICCE63647.2025.10930112

View publication

ICCE 2024

2024

An Efficient Fine-Tuning of Generative Language Model for Aspect-Based Sentiment Analysis

C. Lee, H. Lee, K. Kim, S. Kim, J. Lee

Las Vegas, USA | January 5 - 8, 2024 | doi: 10.1109/ICCE59016.2024.10444216

View publication

Jahr – European Journal of Bioethics

2024

Scaling Up: Analyzing Classification Outcomes in Large Korean Legal Document Datasets

Y.-C. Park, H. Lee, J. Lee

Vol. 15, No. 2, pp. 527-541 | doi: 10.21860/j.15.2.12

View publication

PlatCon 2021

2021

An Efficient Neural Network based on Early Compression of Sparse CT Slice Images

A. Moon, S. Lee, S. Cho, T. Lee, H. Lee, J. Lee

pp. 1-5 | doi: 10.1109/PlatCon53246.2021.9680749

View publication

Featured Build

Project spotlight

A representative system that reflects how I think about AI research, tooling, and reliability.

Featured project

GitHub

AutoLabOS

An operating system for autonomous research

AutoLabOS is a research systems project for running literature-grounded, checkpointed, and inspectable workflows from paper collection to manuscript drafting. Instead of treating research as a single generation step, it structures the process as a governed loop with explicit review, evidence tracking, and resumable state.

TypeScriptNode.jsReactOpenAICodex CLISemantic Scholar

AutoLabOS

Fixed multi-stage workflow for research execution instead of open-ended agent drift

AutoLabOS

Checkpointed runs with inspectable artifacts and resumable progress

AutoLabOS

Evidence-bound claim review that limits conclusions to what the run actually supports

AutoLabOS

Terminal and web interfaces for operating autonomous research pipelines

Journey

Academic timeline

From computer science training to current doctoral research in AI.

  1. 2024.03 - Present

    Ph.D. Course, Department of Artificial Intelligence

    Chung-Ang University

  2. 2022.03 - 2024.02

    M.Sc. Course, Department of Artificial Intelligence

    Chung-Ang University

  3. 2021.09 - 2021.12

    Intern

    S2W Inc.

  4. 2015.03 - 2022.02

    B.Sc. Course, School of Computer Science and Engineering

    Chung-Ang University

  5. 2013.03 - 2015.02

    Hansung Science High School

    Science-specialized high school education

Impact

Awards and selected builds

Recognition and projects that connect academic research with deployable AI systems.

Awards

  • 3rd Prize, 2022 AI Graduate School Challenge, LG
  • 3rd Prize, 2021 Text Ethics Verification Data Hackathon Competition, National Information Society Agency (NIA)

2025 - Present

AutoLabOS

Designed and built an autonomous research system that organizes literature review, hypothesis generation, experiment planning, review gating, and manuscript drafting inside a governed execution loop.

Open-source project

2023.09 - 2024.12

Automatic Generation of Children's Song Lyrics and Improvement of Lyric Quality Based on Large Language Model

Research and development project focused on creative language generation and quality improvement with large language models.

2023.03 - 2024.12

Integrated Framework for Automatic Neural Network Generation and Deployment Optimized for Runtime Environments

Developed an end-to-end framework for generating and deploying neural networks across runtime-constrained environments.

In cooperation with ETRI

Capabilities

Tech stack

Tools I use to move from research ideas to governed experiments, prototypes, and production-facing systems.

Core languages

PythonTypeScriptJavaScriptJavaC++Swift

AI research

PyTorchTensorFlowHugging FaceLangChainLangGraphOpenAIOpenAI AgentsSemantic Scholarscikit-learnWeights and Biases

Systems and deployment

Node.jsNext.jsReactVitestFastAPILinuxmacOSRaspberry Pi

Coverage

In the news

Selected media and university coverage related to my research and collaborations.

Chung-Ang University

NAACL 2025 paper acceptance news from Chung-Ang University

Open

NewsTheAI

Chung-Ang University featured in Safe AI coverage from Singapore

Open

GetNews

LG selected as organizer for the 2022 AI Graduate School Symposium

Open

ETNews

ETRI Invention Camp grand prize coverage

Open

Contact

Let's build something meaningful.

I'm especially interested in trustworthy NLP, autonomous research systems, and collaborations that turn strong ideas into reliable and inspectable AI products.

Emailglhy0718@gmail.comORCID0009-0001-6900-4638Google ScholarHanyong LeeBloglhy0718.github.ioHugging FacelhyX@hanyong_lee_