Young Je Yoon

Integrated M.S.–Ph.D. student, UNIST

My research focuses on characterizing nanoscale structures using X-ray scattering.

My work in large language models investigates how language differences affect information retrieval and reasoning.

Advisor
Prof. Tae Joo Shin
Email
yoonyoungje@unist.ac.kr

Timeline

Years Position Where 2009Present
2023–presentEducation Integrated M.S.–Ph.D. UNIST, Graduate School of Semiconductor Materials and Devices Engineering
2019–2023Industry experience Mass-production technology development Startup development team
2016–2019Industry experience Mass-production technology development Development team, display manufacturer
2009–2016Education B.S., Applied Organic Materials Engineering Inha University

Research

X-ray scattering metrology for semiconductor nanostructures

I study X-ray scattering techniques for the nondestructive characterization of nanoscale semiconductor patterns. My research focuses on critical-dimension small-angle X-ray scattering (CD-SAXS), combining scattering simulations and inverse modeling to reconstruct the cross-sectional profiles of nanoscale patterns. I also investigate metrology challenges associated with EUV lithography and advanced semiconductor fabrication.

Cross-lingual retrieval and LLM evaluation

I study retrieval-augmented language models in cross-lingual settings, where a question in one language requires evidence written in another. My research focuses on multi-hop question answering, with particular attention to retrieval effectiveness and answer faithfulness. I developed K-CrossHop, a Korean–English benchmark for examining where cross-lingual failures arise in retrieval and evidence-based answering.

Simulated CD-SAXS map of a trapezoidal line 60 nm tall, 50 nm wide at the base and 34 nm at the top, repeated every 100 nm. Each column is the Bragg rod of one diffraction order, on a log intensity scale. Rotating the sample through ω = ±60° fills the bow-tie.

Technical expertise

Characterization
X-ray scattering (CD-SAXS), synchrotron beamline instrumentation, SEM, FIB
Nanofabrication
Electron-beam lithography (EBL), inductively coupled plasma reactive-ion etching (ICP-RIE)
Computation
Python, scattering simulation, inverse modeling, numerical optimization, machine learning, LLM evaluation

Publications

  1. 2027
  2. K-CrossHop: A Two-Stage Decomposition of Query-Side Cross-Lingual Failures in Korean–English Agentic RAG

    Young Je Yoon, Tae Joo Shin†

    Findings of the Association for Computational Linguistics: EACL 2027. Accepted, to appear.

Contact

yoonyoungje@unist.ac.kr

Address
UNIST, Ulsan, Republic of Korea