Kichang Yang
Sr. Applied Research Scientist at Kakao Corp
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolKichang Yang is a Senior Applied Research Scientist based in the San Francisco Bay Area with a decade of experience building and optimizing production-scale LLM systems. He led deployment and efficiency improvements for Kakao’s Kanana Flex MoE model—introducing FP8/FP4 training and serving optimizations that cut latency by 40% and sped continual pretraining by 30%—and designed Multi-LoRA serving to enable parallel LoRA deployments without performance loss. His background blends open-source research contributions (EleutherAI’s Oslo and Polyglot projects) with industry impact, including an ACL 2025 Industry Track paper and a featured NVIDIA Developer Blog collaboration. Comfortable across model orchestration, tensor-parallel training, and RLHF dataset pipelines, he pairs deep systems-level engineering with practical production delivery. An avid open-source contributor, he maintains a public GitHub and Google Scholar profile that reflect both research rigor and hands-on ML engineering.
10 years of coding experience
5 years of employment as a software developer
Bachelor's degree School of Software, Bachelor's degree School of Software at Soongsil University
English, Korean
Github Skills (103)
mobile-development10
python10
windows10
huggingface10
deep-learning10
mlops10
javascript10
chatgpt10
app-framework10
linux10
ml10
audio10
llmops10
neural-network10
transformer10
Programming languages (7)
Github contributions (5)
jason9693/Style-Transfer
Feb 2018 - Dec 2018
Stanford CS231n(2017) Assignment3.
Contributions:21 commits, 2 PRs, 19 pushes in 10 months
pytorchstanforddeep-learningstanford-cs231ncs231n
jason9693/midi-neural-processor
Jul 2019 - Jul 2021
midi processor library for PerformanceRNN & MusicTransformer published by "Google Magenta"
Contributions:8 commits, 1 PR, 7 pushes in 2 years
midimagentaaudiomusicpublished
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.