Jang-hyun Kim
Machine Learning Researcher at Apple
Seoul, United States
Summary
👤
Senior🎓
Top SchoolJang-hyun Kim is a machine learning researcher with 8 years of experience, currently working on foundation models at Apple after leading LLM inference-efficiency research as a freelance AI researcher for NAVER Cloud. His work centers on meta-optimization techniques that let trained networks compress datasets and contexts, guide augmentation via saliency, and automatically surface problematic data—reducing reliance on manual curation. He holds a PhD in Computer Science from Seoul National University and has applied deep models to causal discovery in human genes during a visiting scholar stint at NYU. Notably, he led the FastKVZip project on large-scale LLM inference efficiency and has prior industry experience in speech enhancement and applied ad analytics. Based in Seoul with international research exposure, he combines theoretical rigor with practical system-focused contributions that speed training and inference.
8 years of coding experience
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Seoul National UniversitySeoul Science High School
English
Github Skills (36)
speaker10
verification10
loss10
speaker-verification9
puzzle9
robustness9
image-matching9
compression9
outlier-detection8
statistics8
novelty-detection8
anomaly-detection8
pytorch7
benchmark7
llm7
Programming languages (2)
Github contributions (5)
Janghyun1230/Speaker_Verification
Dec 2019 - Oct 2021
Tensorflow implementation of generalized end-to-end loss for speaker verification
Contributions:11 commits, 5 PRs, 28 pushes in 1 year 10 months
end-to-endspeakerspeaker-verificationlosstensorflow
Janghyun1230/SpeechEnhancement
Jun 2020 - Oct 2021
Contributions:13 commits, 11 pushes, 1 branch in 1 year 3 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.