cv

Teaching Assistant - AI Application Specialist

Samsung Electronics Co.

Delivered sessions on language model applications and Agentic AI to software developers.

• Language Model Applications
• Agentic AI Systems
• Enterprise AI Integration

Teaching Assistant - Level 4 Data Science Expert Program

Samsung Electronics Co.

Led hands-on sessions on Agentic AI for data science professionals, with 90% of participants holding Ph.D. degrees.

• Advanced Agentic AI Techniques
• Ph.D.-level Training
• Hands-on Workshops

Research Intern

Oracle Labs

Developed NL2SQL generation model for Oracle Database, achieving 27.4% improvement in execution accuracy over baseline.

• NL2SQL Generation
• Oracle Database Optimization
• 27.4% Performance Improvement

Research Intern

Samsung Electronics' Mobile Experience Business

Developed image-processing algorithms for mobile applications.

• Image Processing
• Mobile Algorithms
• Computer Vision

Research Intern

AMI Laboratory

Developed an end-to-end method for reconstructing animatable 3D human mesh from a single cropped 2D image of a person.

• 3D Human Reconstruction
• Computer Vision
• Grand Prize Winner

Product Manager Intern

Sellerhub

Served as a product manager at Sellerhub, a startup providing a platform for managing online shopping mall integrations.

• Product Management
• E-commerce Platform
• Startup Experience

Ph.D.

Pohang University of Science and Technology (POSTECH)

Artificial Intelligence

• Neural Information Retrieval
• Multi-modal Question Answering
• Agentic AI

Bachelor

Pohang University of Science and Technology (POSTECH)

Electrical Engineering

• Electrical Engineering
• Computer Science

High School Diploma

Chungnam Science High School

Science

• Advanced Science Curriculum

KDD Cup Meta CRAG 2024 - Multiple First Places

KDD Cup Meta CRAG Challenge

First place in comparison questions (Tasks 1, 2, 3) and post-processing question (Task 1) as head member of team dRAGonRAnGers.

Newcomb Lim Ki-Hong Design Challenge Grand Prize

POSTECH

Awarded to students of POSTECH EE who presented outstanding research results in a design challenge.

Millitech Military Academy Commander's Award

Millitech Military Academy

Awarded to a ROND (Benchmarking of Israel's Talpiot) novice who showed outstanding research results.

National Science and Technology Scholarship

Korean Government

Scholarship awarded to students with outstanding academic ability in the field of science and engineering.

HELIOS: Harmonizing Early Fusion, Late Fusion, and LLM Reasoning for Multi-Granular Table-Text Retrieval

ACL 2025 Main

HELIOS formulates retrieval as finding a query-relevant subgraph within a bipartite data graph built via early fusion of table segments and passages, and introduces a three-stage pipeline integrating early fusion, late fusion, and LLM reasoning.

SAFE: Schema-Driven Approximate Distance Join for Efficient Knowledge Graph Querying

EMNLP 2025

SAFE introduces a schema-driven approximate distance join algorithm that refines noisy LLM-generated query graphs using schema-level constraints and efficiently aligns them with large knowledge graphs.

SPARTA: Scalable and Principled Benchmark of Tree-Structured Multi-hop QA over Text and Tables

Submitted

SPARTA is a fully automated SQL-centric pipeline that constructs a tree-structured multi-hop QA benchmark by unifying structured and unstructured evidence from tables and text into a single relational representation.

KDD Cup Meta CRAG 2024 Technical Report: Three-step Question-Answering Framework

2024 KDD Cup Workshop for Retrieval Augmented Generation

A three-step RAG framework that minimizes unnecessary retrievals by leveraging LLMs' inherent knowledge and introduces a verification stage to prevent error propagation.

HELIOS: Multi-Granular Table-Text Retrieval

Harmonizing Early Fusion, Late Fusion, and LLM Reasoning for advanced retrieval systems that handle both structured and unstructured data.

• ACL 2025 Main
• SOTA Performance
• Multi-modal Retrieval

KDD Cup Meta CRAG 2024

Three-step Question-Answering Framework for Agentic AI that minimizes unnecessary retrievals and prevents error propagation.

• First Place Winner
• Agentic AI Framework
• Error Prevention

Oracle NL2SQL Generation

Development of NL2SQL generation model specifically optimized for Oracle Database systems with significant performance improvements.

• 27.4% Improvement
• Oracle Database
• SQL Generation