SP.

Seoul, Republic of Korea

I build AI systems that make complex knowledge useful.

I’m Suhyeong Park, an AI researcher exploring natural language processing, retrieval-augmented generation, and reliable LLM agents.

Explore my research

01 / About

Research with a path to real-world impact.

I work broadly across NLP and LLM systems, with an occasional deep dive into BioNLP.

I develop broadly applicable methods for retrieval, knowledge grounding, and LLM agents. When a problem calls for domain expertise, I bring those ideas into BioNLP settings such as medical consultation, clinical-trial tracking, and pediatric emergency support.

I am completing a B.S. in Data Science at The Catholic University of Korea and will join Korea University’s Data Mining & Information Systems Lab through the integrated M.S./Ph.D. program in Computer Science under Prof. Jaewoo Kang.

1Publication
4×1st-prize awards
2022—26B.S. in Data Science
3Research environments

02 / Publication

Selected research.

EMNLP 2025 Suzhou, China pp. 30447–30463

Leveraging Knowledge Graph-Enhanced LLMs for Context-Aware Medical Consultation

Su-Hyeong Park*, Ho-Beom Kim*, Seong-Jin Park, Dinara Aliyeva, and Kang-Min Kim†

* Equal contribution† Corresponding author

A context-aware medical consultation approach that combines structured knowledge graphs with large language models.

Read on Paper Code

03 / Experience

A timeline of learning by building.

2025.07 — Present

AIGEN Sciences · Seongsu

AI Research Intern

  • Developing an LLM agent for clinical-trial research tracking.
  • Building a pediatric emergency retrieval-augmented chatbot system.
2024.06 — 2025.07

Biomedical Acoustics Innovation Laboratory

Undergraduate Researcher

  • Studied ultrasound analysis through signal and image processing.
  • Participated in research paper seminars.
2022.05 — 2024.06

Medical AI Lab

Undergraduate Researcher

  • Worked on clinical text analysis using natural language processing.
  • Participated in research paper seminars.

04 / Selected work

Projects that shaped how I think.

01 Reinforcement Learning · 2025 Efficient Lesion Search in Ultrasound Images via RL

Trained a PPO agent to learn efficient ultrasound scan paths for lesion localization.

PPOPyTorchMedical Imaging
02 Natural Language Processing · 2024 LLM Educational Chatbot with Celebrity Persona

Combined web-grounded persona generation with semantic retrieval over curated coding problems.

RAGLLMVector DB
View on GitHub
03 Retrieval Systems · 2024 MPRAG: Momentum-Precise RAG for Open-Domain QA

Proposed momentum-based embedding alignment and graph reranking with Monte Carlo search.

RAGGraph RerankingQA
View on GitHub
04 Deep Learning · 2024 Named Entity Recognition for Kiosk Interaction

Built a RoBERTa multi-task model for store classification and entity-aware ordering.

RoBERTaNERMulti-task

05 / More

Recognition & community.

Awards

  1. 2024

    1st Prize, Biohealth Challenge
    Signal-based ML for breast-cancer detection using Doppler ultrasound.

  2. 2023

    1st Prize, Data Analysis Competition
    Deep-learning anomaly detection for PCB images.

  3. 2023

    1st Prize, Medical AI Track
    Whole-slide image classification with multiple-instance learning.

  4. 2022

    1st Prize, Data Analysis Competition
    Pet disease classification from structured veterinary data.

Activities

  1. 2024

    NAVER AI RUSH Ambassador
    Built an educational RAG chatbot with HyperCLOVA X.

  2. 2023—24

    Executive Member, ToBig’s
    Led NLP paper reviews and represented the club at a U.S. Embassy Tech Talk.

Toolkit

PythonPyTorchHugging FaceLangChainLangGraphTensorFlowScikit-learnPandasC/C++