United States
Director of Language & Knowledge Lab.
LBox is a legal tech startup in Korea that develops data analytics platform for lawyers. In 2020, I developed the machine learning toolset (LBox ML) and designed MLOps pipeline on AWS SageMaker. Since 2021, my role has shifted to research advising and recruiting top talents.
I led Twelve Labs' research team which is responsible for developing AI engines (Pegasus, Marengo) that power the company's core products.
Twelve Labs and KAIST jointly formed a team to participate in a video search competition held by Microsoft at ICCV 2021 (VALUE Challenge) and placed the first place in Retrieval Phase, outperforming other competent teams including Tencent, Columbia University, and Kakao Brain. As a startup+school team without large GPU clusters, we used more than USD 50,000 on AWS EC2 (Cloud Computing) in 6 weeks to rapidly scale up the experiments and thus be able to compete with the large enterprises.
Worked on semiparametric language models.