PeopleofSuperintelligence

Researcher

Ronghang Hu

Also known as 胡戎航

Ronghang Hu is a computer vision researcher who works on models that connect language, images and video, and who was a core contributor to the SAM 2 and SAM 3 segmentation models. He earned a B.Eng. at Tsinghua University in 2015 and a Ph.D. in computer science at UC Berkeley in 2020, with papers in 2016 and 2017 on language-guided segmentation, module networks for visual question answering and Learning to Segment Every Thing. At Meta FAIR from 2020 to 2025, he coauthored the FLAVA vision-and-language model in 2021, and SAM 2 received an Outstanding Paper Honorable Mention at ICLR 2025. Since November 2025, he is a member of technical staff at SpaceX Super Intelligence in Palo Alto, where he works on multimodal AI.

Now at
SpaceX
Badges

Timeline

  1. 2011–2020

    Tsinghua and a Berkeley PhD

    He studies at Tsinghua University from 2011 and earns a B.Eng. in Electronic Information Science and Technology in 2015. He then completes a Ph.D. in Computer Science at the University of California, Berkeley in 2020.

  2. 2016–2017

    Papers on segmentation and reasoning

    In 2016, he is first author of Segmentation from Natural Language Expressions, written with Marcus Rohrbach and Trevor Darrell, which segments an image from a phrase such as "two men sitting on the right bench". In 2017, he is first author of Learning to Reason: End-to-End Module Networks for Visual Question Answering, written with Jacob Andreas, Rohrbach, Darrell and Kate Saenko, whose networks predict instance-specific layouts without a parser. In November 2017, he is also first author of Learning to Segment Every Thing, written with Piotr Dollár, Kaiming He, Darrell and Ross Girshick, which trains Mask R-CNN to detect and segment 3000 visual concepts.

  3. 2020–2025

    Research scientist at Meta FAIR

    After research internships at Facebook AI Research in 2017 and 2019, he joins Meta FAIR in Menlo Park as a research scientist in June 2020. He stays until November 2025.

  4. 2021–2022

    The FLAVA foundation model

    In December 2021, FLAVA: A Foundational Language And Vision Alignment Model appears, with him as one of three authors with equal contribution, alongside Amanpreet Singh and Vedanuj Goswami. The single model performs well across 35 vision, language and vision-and-language tasks and is presented at CVPR 2022.

  5. 2024

    SAM 2 for images and video

    In August 2024, SAM 2: Segment Anything in Images and Videos appears with 18 authors, and he is one of its core contributors. The model uses 3x fewer interactions than prior approaches in video segmentation and is more accurate and 6x faster than the Segment Anything Model on images.

  6. 2025

    An ICLR honorable mention

    SAM 2 receives one of three Honorable Mentions at the ICLR 2025 Outstanding Paper Awards, with Hu among its authors.

  7. November 2025

    SAM 3 and concept prompts

    In November 2025, SAM 3: Segment Anything with Concepts appears, and he is a core contributor. The model detects, segments and tracks objects in images and videos from short noun phrases or image exemplars, and doubles the accuracy of existing systems on promptable concept segmentation.

  8. Since November 2025

    Technical staff at SpaceX

    He becomes a member of technical staff at SpaceX Super Intelligence in Palo Alto in November 2025, working on multimodal AI.

Career

  1. Since 2025SpaceXMember of Technical Staff
  2. 2017–2025FAIR (Meta)Research Intern, Research Scientist

Education

  1. 2020University of California, BerkeleyPh.D.
  2. 2015Tsinghua UniversityB.Eng.

From the biography.

  1. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Meta. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  2. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Meta. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  3. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Meta. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  4. 22 of 133 people shaping AI with a former OpenAI role now work at AnthropicIncludes Meta. Our estimates of where OpenAI recruits came from and where 133 people with former OpenAI roles work now, including the companies and labs they founded.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. Ronghang Hu – Advancing multimodal perception, understanding, and reasoning ronghanghu.com. Checked 22 October 2026.
  2. [1603.06180] Segmentation from Natural Language Expressions arxiv.org. Checked 22 October 2026.
  3. [1704.05526] Learning to Reason: End-to-End Module Networks for Visual Question Answering arxiv.org. Checked 22 October 2026.
  4. [1711.10370] Learning to Segment Every Thing arxiv.org. Checked 22 October 2026.
  5. [2112.04482] FLAVA: A Foundational Language And Vision Alignment Model arxiv.org. Checked 22 October 2026.
  6. [2408.00714] SAM 2: Segment Anything in Images and Videos arxiv.org. Checked 22 October 2026.
  7. Announcing the Outstanding Paper Awards at ICLR 2025 – ICLR Blog blog.iclr.cc. Checked 22 October 2026.
  8. [2511.16719] SAM 3: Segment Anything with Concepts arxiv.org. Checked 22 October 2026.