PeopleofSuperintelligence

Researcher

Ji Lin

Ji Lin is an AI researcher whose work spans efficient computer vision, TinyML, low-bit language models, and visual reasoning. He coauthored TSM for video understanding in 2019 and MCUNet for ImageNet-scale inference on microcontrollers in 2020. After earning degrees at Tsinghua and MIT, he completed an MIT EECS PhD under Song Han in 2023 and joined OpenAI as a research scientist, contributing to multimodal and reasoning systems including Operator. He received a Qualcomm Innovation Fellowship in 2022, coauthored AWQ and VILA in 2024, and the AWQ team won the MLSys 2024 Best Paper Award. He joined Meta's superintelligence team in 2025 and works in its TBD lab as of October 2026.

Now at
Meta
Works
50
Citations
9,895
h-index
30
Badges

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2019

    TSM for video understanding

    In 2019, he coauthors TSM, which shifts part of a 2D CNN's channels across video frames for temporal modeling with zero extra computation and zero parameters.

  2. 2020

    MCUNet on microcontrollers

    In 2020, he coauthors MCUNet, which pairs TinyNAS, a neural architecture search method, with TinyEngine, a lightweight inference engine, for ImageNet-scale inference on microcontrollers.

  3. 2022

    A Qualcomm fellowship

    He receives a Qualcomm Innovation Fellowship in 2022.

  4. 2023

    MIT EECS doctorate

    He completes a PhD in MIT EECS in December 2023 under Song Han. His earlier degrees are a B.Eng. in electronic engineering from Tsinghua and an M.Sc. in EECS from MIT.

  5. 2023–2025

    OpenAI research scientist

    He joins OpenAI as a research scientist in 2023 and works on multimodal, reasoning, and synthetic-data systems. He contributes to Operator's reasoning stack.

  6. 2024

    AWQ and an MLSys award

    He coauthors AWQ in 2024, a hardware-friendly method for low-bit LLM weight quantization that protects salient weights from quantization error. The AWQ team receives the MLSys 2024 Best Paper Award.

  7. 2024

    VILA for visual language

    He coauthors VILA in 2024, a visual language model family built with an enhanced pretraining recipe. VILA supports multi-image reasoning and on-device deployment on Jetson Orin.

  8. Since 2025

    Meta's TBD lab

    In June 2025, he joins Meta's superintelligence team. He works in Meta's TBD lab as of October 2026.

Career

  1. Since 2025Meta
  2. 2023–2025OpenAIResearch scientist

Education

  1. 2023Massachusetts Institute of TechnologyM.Sc.
  2. Tsinghua UniversityB.Eng.

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. TSM: Temporal Shift Module for Efficient Video Understanding2019 · IEEE/CVF International Conference on Computer Vision (ICCV) · 2,132 citations
  2. AMC: AutoML for Model Compression and Acceleration on Mobile Devices2018 · Lecture notes in computer science · 1,399 citationsWith Li-Jia Li
  3. HAQ: Hardware-Aware Automated Quantization With Mixed Precision2019 · IEEE Computer Society Conference on Computer Vision and Pattern Recognition · 972 citations
  4. Searching Efficient 3D Architectures with Sparse Point-Voxel Convolution2020 · Lecture notes in computer science · 780 citations
  5. Differentiable Augmentation for Data-Efficient GAN Training2020 · 395 citations

From OpenAlex, as of 21 October 2026.

  1. 22 of 133 people shaping AI with a former OpenAI role now work at AnthropicIncludes Ji Lin. 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.
  2. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Ji Lin. 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.
  3. 16% of AI career moves in the 2020s went from a university to a companyIncludes Ji Lin. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  4. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Ji Lin. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.

Colleagues

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

Sources and credits

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

  1. TSM: Temporal Shift Module for Efficient Video Understanding openaccess.thecvf.com. Checked 21 October 2026.
  2. MCUNet: Tiny Deep Learning on IoT Devices papers.nips.cc. Checked 21 October 2026.
  3. System brings deep learning to “internet of things” devices news.mit.edu. Checked 21 October 2026.
  4. Ji Lin - MIT HAN Lab hanlab.mit.edu. Checked 21 October 2026.
  5. Ji Lin's Homepage linji.me. Checked 21 October 2026.
  6. Here Is Everyone Mark Zuckerberg Has Hired So Far for Meta’s ‘Superintelligence’ Team wired.com. Checked 21 October 2026.
  7. AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration hanlab.mit.edu. Checked 21 October 2026.
  8. VILA: On Pre-training for Visual Language Models openaccess.thecvf.com. Checked 21 October 2026.
  9. MIT HAN Lab hanlab.mit.edu. Checked 21 October 2026.