Portfolio

Junyeong Song

Computer Science student bridging AI research and production โ€” one rebuild at a time.

Hi, I'm Junyeong Song.

I don't just use AI tools โ€” I rebuild them. Diffusion models, agent harnesses, RAG pipelines, OCR benchmarks: I build minimal versions from scratch because that's the fastest way to know exactly where they break and how to fix them.

I'm especially drawn to agentic systems โ€” how LLMs reason, plan, and coordinate โ€” and the gap between what works in a paper and what holds up in practice.

What drives me is the moment a stripped-down prototype teaches me something a paper didn't. Build small, learn fast, then make it real.

Junyeong Song
UNIST ยท Computer Science and Engineering

01Education

UNIST

Computer Science and Engineering

2020.03 - Present

Military Service

Under Military Service

2022.07 - 2024.01

Daegu il Science High School

Physics & Computer Science

2017.03 - 2020.02

02Research Interests

Synthetic Data Generation Computer Vision Vision LLM Diffusion Models

03Publications

๐Ÿ“‘

KorMedMCQA

Multi-Choice Question Answering Benchmark for Korean Healthcare Professional Licensing Examinations

arXiv โ†’

04Experiences


05Projects


06Awards & Others

๐Ÿฅ‡

Awards

Graduated 1st and entered UNIST with Unistar Scholarships

๐Ÿ‘จโ€๐Ÿซ

Teaching

TA in AI Intro and Programming 1