Skip to content
View kbannie's full-sized avatar
🎯
Focusing
🎯
Focusing
16 followers · 18 following

Achievements

Achievement: Pull Sharkx2

Achievements

Achievement: Pull Sharkx2

Organizations

@Backend-graduate @Dukvis @Corner-1th-Hackathon @DS-laver @jinakayoo @ezpz4 @Weather-of-Ingredients

Block or report kbannie

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
kbannie/README.md

Kabeen Kim — Document AI, GraphRAG, and LLM Reasoning

Website  ·  Google Scholar  ·  LinkedIn  ·  CV  ·  Email

I'm Kabeen, an undergraduate studying Computer Engineering and Information Statistics at Duksung Women's University. I work on document understanding, retrieval-augmented generation, and LLM reasoning.

I care about how models fail, which evidence they retrieve, and how to design experiments that explain their behavior.

  • Recent focus: multi-hop GraphRAG over knowledge graphs, including path-preserving search and evidence reranking at ETRI.
  • Highlight: 3rd place out of 210 teams, 2025 Samsung AI Challenge — Visually-Rich Document Understanding, with CoDU.

Selected research

UDAPose

COMPUTER VISION · CVPR 2026

Unsupervised domain adaptation for human pose estimation in low-light images.

Haopeng Chen, Yihao Ai, Kabeen Kim, Robby T. Tan, Yixin Chen, Bo Wang

Paper →   Official code →

ConFuse

LLM + TABULAR LEARNING · ICTC 2026, ACCEPTED

Context-aware fusion of GBDT and fine-tuned LLM predictions for sleep prediction from lifelog data.

Kabeen Kim*, Yena Kim*, Minjeong Seo*  (*equal contribution)

Code & experiments →

CoDU

DOCUMENT UNDERSTANDING · KDBC 2025

Multi-stage document parsing with layout detection, OCR, formula recognition, and reading-order reconstruction.

🏆 3rd / 210 teams, 2025 Samsung AI Challenge

Kabeen Kim, Minhye Lee, Haein Seo, Jehyeok Rew

Code & models →

SPTC

LLM REASONING · KCC 2025

A single-pass tree-based prompting scheme for efficient reasoning with small language models.

Kabeen Kim, Jiye Park, Jehyeok Rew

Paper →

Applied AI: DS Chatbot — a RAG assistant for university regulations, built with LangChain, Pinecone, and Streamlit.

Experience

Organization Role Focus
ETRI Research Intern
Jul – Aug 2026		 Knowledge graphs & multi-hop GraphRAG
Boston Consulting Group Research Analyst Intern
Mar – May 2026		 Industrial data analysis & LLM workflows
DMKD Lab, Duksung Women's University Undergraduate Researcher
Mar 2025 – Feb 2026		 LLM reasoning & document understanding
University of Mississippi Undergraduate Researcher
Aug 2024 – Feb 2025		 Low-light human pose estimation

Full timeline and publications: academic homepage.

Toolkit

Area Technologies
Languages & data Python · SQL · C++ · Pandas · NumPy
Machine learning PyTorch · Hugging Face Transformers · LoRA / QLoRA · LightGBM · XGBoost
Retrieval & documents LangChain · Pinecone · DocLayout-YOLO · PaddleOCR · Pix2Tex
Experiments & infrastructure Weights & Biases · Docker · AWS EC2 / S3 · Linux · Git

Pinned Loading

  1. CoDU CoDU Public

    Complex Document Understanding: layout detection, OCR, formula recognition, and reading-order reconstruction.

    Python

  2. ConFuse ConFuse Public

    ConFuse: Context-Aware Fusion of LLMs and GBDTs for Sleep Prediction from Lifelog Data

    Python