PeopleofSuperintelligence

Researcher

Miao Liu

Also known as 刘淼

Miao Liu researches first-person vision, augmented reality, and generative AI models. They earned a bachelor's degree from Beihang University in 2015, a master's degree from Carnegie Mellon University in 2017, and a Ph.D. from Georgia Tech in 2022. At Meta GenAI, they moved from research scientist to senior research scientist and contributed to Emu, Llama 3, and the multimodal part of Llama 4. Their work includes Ego4D and a transformer paper on egocentric gaze; the LEGO action-frame paper became an ECCV 2024 award candidate and an EgoVis distinguished paper. In August 2025, they joined Tsinghua's College of AI as an assistant professor and lead the MEOW Lab; in 2026, they coauthored IPIBench, a benchmark for proactive intelligence in multimodal LLMs under streaming video.

Badges

Timeline

  1. 2015–2017

    Degrees at Beihang and Carnegie Mellon

    They earn a bachelor's degree from Beihang University in 2015 and a master's degree from Carnegie Mellon University in 2017.

  2. 2021

    Ego4D and first-person video

    In 2021, they coauthor Ego4D, a massive-scale egocentric video dataset and benchmark that introduces 3,670 hours of daily-life footage.

  3. 2022

    A Ph.D. and a gaze paper

    They complete a Ph.D. at Georgia Tech in June 2022 and coauthor the 2022 paper “In the Eye of Transformer: Global-Local Correlation for Egocentric Gaze Estimation.”

  4. August 2022–August 2025

    Research at Meta GenAI

    From August 2022 to June 2024, they serve as a research scientist at Meta GenAI; from June 2024 to August 2025, they serve as a senior research scientist there. During the Meta GenAI period, they contribute to the development of Emu, Llama 3, and the multimodal part of Llama 4.

  5. November 2022

    A BMVC Best Student Paper Award

    In November 2022, they and their coauthors win BMVC's Best Student Paper Award for “In the Eye of Transformer.”

  6. October 2024–June 2025

    LEGO and EgoVis recognition

    At ECCV 2024, they coauthor LEGO: Learning EGOcentric Action Frame Generation via Visual Instruction Tuning. In October 2024, the paper is named among 15 ECCV award candidates; in June 2025, it appears among the EgoVis 2023/2024 Distinguished Paper winners.

  7. Since 2025

    Assistant professor at Tsinghua

    In August 2025, they join Tsinghua's College of AI as an assistant professor. As of October 2026, they lead the MEOW Lab and research first-person vision, augmented reality, and generative AI models. In 2026, they coauthor IPIBench, a benchmark for interactive proactive intelligence in multimodal LLMs under streaming video.

Career

  1. Since 2025Tsinghua UniversityAssistant Professor
  2. 2022–2025MetaResearch Scientist, Senior Research Scientist

Education

  1. 2022Georgia Institute of TechnologyPhD
  2. 2017Carnegie Mellon UniversityMaster's degree
  3. 2015Beihang UniversityBachelor's degree

From the biography.

Colleagues

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

Sources and credits

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

  1. Miao LIU-College of AI, Tsinghua University collegeai.tsinghua.edu.cn. Checked 26 October 2026.
  2. Ego4D: Around the World in 3,000 Hours of Egocentric Video arxiv.org. Checked 26 October 2026.
  3. In the Eye of Transformer: Global-Local Correlation for Egocentric Gaze Estimation arxiv.org. Checked 26 October 2026.
  4. Miao Liu 刘淼 | Tsinghua University · MEOW Lab aptx4869lm.github.io. Checked 26 October 2026.
  5. The 33rd British Machine Vision Conference 2022: Paper Awards bmvc2022.cvml.group. Checked 26 October 2026.
  6. EgoVis 2023/2024 Distinguished Paper Awards egovis.github.io. Checked 26 October 2026.
  7. [2605.27074] IPIBench: Evaluating Interactive Proactive Intelligence of MLLMs under Continuous Streams arxiv.org. Checked 26 October 2026.