Hwanjun Song
About Hwanjun Song
Hwanjun Song is an Assistant Professor at KAIST, specializing in evaluating outputs of large language models and text summarization. He has transitioned from roles at major tech companies like Amazon Web Services and Google to an academic position since 2023.
Work at KAIST
Hwanjun Song currently holds the position of Assistant Professor at the Korea Advanced Institute of Science and Technology (KAIST) in Daejeon, South Korea. He has been in this role since 2023, contributing to the academic community through research and teaching. His work focuses on evaluating outputs of large language models, particularly in text summarization. This position marks a transition from industry roles to academia, where he applies his extensive knowledge and experience in data science.
Experience at Amazon Web Services
Before joining KAIST, Hwanjun Song worked at Amazon Web Services (AWS) as a Research Scientist for a period of nine months in 2023. His role was based in Bellevue, Washington, United States. During his time at AWS, he engaged in research that contributed to advancements in technology and data science, leveraging his expertise in large language models.
Background in Data Science Education
Hwanjun Song earned his Doctor of Philosophy (PhD) in Data Science from the Korea Advanced Institute of Science and Technology (KAIST). His studies spanned from 2016 to 2021, during which he developed a strong foundation in data science methodologies and applications. This educational background has equipped him with the skills necessary for both research and teaching in his current academic role.
Previous Roles in Major Tech Companies
Prior to his current position, Hwanjun Song gained valuable experience in the tech industry. He worked as a Research Scientist at NAVER Corp from 2021 to 2023 in Gyeonggi, South Korea. Additionally, he completed a five-month internship as a Research Intern at Google in 2020 in Mountain View, California, United States. These roles allowed him to refine his research skills and contribute to significant projects in the field of data science.
Research Focus and Contributions
Hwanjun Song's research primarily concentrates on improving the efficiency of Transformer models, specifically in their inference processes. He investigates attention mechanisms and auto-regressive decoding to enhance the performance of large language models. His work aims to advance the understanding and application of these technologies in various domains, particularly in text summarization.