Kim, Jaedeok
Sr. DevTech Engineer, NVIDIA
Ph.D., Department of Mathematical Sciences, KAIST
Contact : 05jaedeok_at_kaist.ac.kr
Research Interests
My research interests lie in developing tools of artificial intelligent system which are useful from the industrial viewpoint. The pursuing research objective is to develop practical and meaningful techniques in practice, which are based on deep learning and mathematics. My recent interests is in optimizing the performance of large models to make them more accessible and cost-effective. This endeavor aims at enabling a broader range of companies to harness the advantages of the AI era. Previously, my work centered on designing lightweight and efficient AI applications, often referred to as on-device AI.
Work Experiences
Senior AI Developer Technology Engineer (April 2021 - present)
AI engineer responsible for developing technologies to efficiently support AI applications on NVIDIA GPU.
Develop frameworks to significantly accelerate inference of large language models: FasterTransformer and TensorRT-LLM.
Staff Engineer (December 2017 - March 2021)
AI Center, Samsung Research.
Lead researcher and software engineer. Solved the problems in developing and operating AI services and had contributed in a way that we provided algorithms or tools to our partners in Samsung - automation algorithm and tool to compress neural networks for on-device AI or data refining tools to reduce the cost of AI service pipelines, etc.
Visiting Researcher ( November 2019 - April 2020)
CILVR, Center for Data Science, New York University
Researcher responsible for collaboration to solve language modeling research projects and patenting new inventions. I focused on modeling fast and memory efficient neural machine translation models and solving the inconsistency issue of incomplete decoding in generation tasks.
Senior Researcher (September 2015 - November 2017)
AI Team, Software R&D Center, Samsung Electronics.
Researcher responsible for solving the issues that need the mathematical foundation including machine learning and data analysis. I had done many AI projects with broad business domains including smart phones, home appliances, or networks.
Educations
Ph. D. (February 2011 - August 2015)
Department of Mathematical Sciences, KAIST
Thesis: Stochastic Modeling and Analysis for Performance Optimization in Wireless Ad hoc Networks
Advisor: Prof. Hwang, Ganguk
M.S. (February 2009 - January 2011)
Department of Mathematical Sciences, KAIST
Advisor: Prof. Hwang, Ganguk
B.S. (March 2005 - February 2009)
Department of Mathematics (major), KAIST
Department of Computer Sciences (minor), KAIST
Awards and Honors
Samsung Best Paper Awards (2019)
Samsung Scholarship (Sep 2014 - Aug 2015)
National Scholarship (Feb 2009 - Aug 2014)
National Research Foundation of Korea (Mar 2005 - Jan 2009)
Shortlist of Future Basic Scientist Fellowship (Top 6 of math department) (2012)