PeopleofSuperintelligence

Researcher

Hwanjun Song

Hwanjun Song is a KAIST researcher whose work spans data-centric learning, language-model evaluation and AI safety. He interned at Google Research in 2020, earned a KAIST PhD in 2021, and worked at NAVER and Amazon Web Services. His work includes a noisy-label survey, the FineSurE summary evaluator, and FeedSum and SummLlama3-8b, which use LLM feedback to improve summaries. As of October 2026, he is an assistant professor and lab director at KAIST, where his research spans human-AI alignment and his group develops trustworthy agentic AI.

Now at
KAIST
Badges

Timeline

  1. 2020–2021

    A Google internship and PhD

    He interns at Google Research in 2020. In February 2021, he earns a PhD from KAIST's Graduate School of Knowledge Service Engineering, now the Graduate School of Data Science.

  2. 2021–2022

    NAVER and noisy-label research

    He works as a research scientist at NAVER in 2021–2022. In March 2022, his survey, Learning from Noisy Labels with Deep Neural Networks: A Survey, appears in IEEE Transactions on Neural Networks and Learning Systems and reviews 62 methods for robust training.

  3. 2023

    Research at Amazon Web Services

    He works as a research scientist at Amazon Web Services in 2023.

  4. 2024

    Joining KAIST's faculty

    In 2024, he joins KAIST as an assistant professor in the Graduate School of Data Science and Department of Industrial and Systems Engineering. He receives a silver prize at Samsung's Humantech Paper Awards for signal processing and NLP in 2024.

  5. August 2024

    FineSurE evaluates summaries

    In 2024, he coauthors FineSurE, an LLM evaluator that assesses summaries for faithfulness, completeness and conciseness.

  6. April 2025

    LLM feedback for summaries

    In April 2025, he coauthors FeedSum, a dataset of multi-dimensional LLM feedback on summaries, and SummLlama3-8b, which generates human-preferred summaries better than the nearly 10-times-larger Llama3-70b-instruct.

  7. July 2026

    A data-centric alignment survey

    In July 2026, he publishes a solo-authored survey that frames alignment tuning for large language models as a data-pipeline design problem.

  8. As of October 2026

    AI safety and lab leadership

    As of October 2026, he is an assistant professor in KAIST's Graduate School of Data Science and Department of Industrial and Systems Engineering, an adjunct professor in the KAIST College of AI, and director of the Data Intelligence System Lab. His research spans AI safety, trustworthy AI and human-AI alignment, and his lab develops data-centric agentic AI systems.

Career

  1. Since 2024KAISTAssistant Professor
  2. 2023AmazonResearch Scientist
  3. 2021–2022NaverResearch Scientist
  4. 2020Google ResearchIntern

Education

  1. 2021KAISTPhD

From the biography.

  1. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  2. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Google. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Hwanjun Song yet. Send us a free-licensed photo.

  1. Hwanjun Song - Korea Advanced Institute of Science and Technology pure.kaist.ac.kr. Checked 21 October 2026.
  2. Hwanjun Song songhwanjun.github.io. Checked 21 October 2026.
  3. [2007.08199] Learning from Noisy Labels with Deep Neural Networks: A Survey arxiv.org. Checked 21 October 2026.
  4. FineSurE: Fine-grained Summarization Evaluation using LLMs - ACL Anthology aclanthology.org. Checked 21 October 2026.
  5. Learning to Summarize from LLM-generated Feedback - ACL Anthology aclanthology.org. Checked 21 October 2026.
  6. Alignment Tuning for Large Language Models: A Data-Centric Lens on Alignment Data Pipelines - ACL Anthology aclanthology.org. Checked 21 October 2026.
  7. Faculty | AX aicollege.kaist.ac.kr. Checked 21 October 2026.
  8. Members | Data Intelligence System Lab disl-lab.github.io. Checked 21 October 2026.
  9. Data Intelligence System Lab (DISLab) disl-lab.github.io. Checked 21 October 2026.