PeopleofSuperintelligence

Researcher

Min Lin

Min Lin is a machine-learning researcher whose work spans convolutional networks, online continual learning and scientific computing. He coauthored Network In Network in 2013 and later studied how learning systems can adapt to a stream of data. At Sea AI Lab, he coauthored work on vision-language security and the Stochastic Taylor Derivative Estimator, which won the 2024 NeurIPS Best Paper Award. In 2025, he examined R1-Zero-like training and explained how JAX supports higher-order derivatives for scientific computing. As of October 2026, he is a founding scientist at AMI Labs and heads its Singapore team.

Now at
AMI Labs

Timeline

  1. 2013

    Network in Network

    In 2013, he coauthors Network In Network with Qiang Chen and Shuicheng Yan. The paper uses micro neural networks built from multilayer perceptrons and applies global average pooling for image classification.

  2. 2019–2020

    Online continual learning

    In 2019, he coauthors work on online continual learning with no task boundaries. A 2020 paper he coauthors proposes Continual-MAML, an online baseline for agents to adapt to new tasks while retaining memory of earlier tasks.

  3. 2023

    Vision-language model robustness

    In 2023, he coauthors a NeurIPS study on adversarial robustness in large vision-language models. It tests black-box attacks on pretrained CLIP and BLIP models, then transfers those attacks to other systems.

  4. 2024

    Stochastic Taylor Derivative Estimator

    In 2024, he coauthors the Stochastic Taylor Derivative Estimator (STDE), which uses high-order automatic differentiation to estimate arbitrary differential operators. The paper wins the NeurIPS Best Paper Award.

  5. February 2025

    Testing an R1-Zero training pattern

    In February 2025, he coauthors a pilot study that examines R1-Zero-like training and the “Aha moment” reported during reinforcement learning.

  6. September 2025

    JAX for scientific computing

    In September 2025, he and Zekun Shi discuss JAX’s Taylor-mode automatic differentiation for physics-informed learning. They explain how STDE uses JAX to estimate arbitrary differential operators efficiently.

  7. As of October 2026

    Head of AMI Labs Singapore

    As of October 2026, he is a founding scientist at AMI Labs and heads its Singapore team, where he leads model-training systems and infrastructure.

  1. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes AMI Labs. 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.
  2. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes AMI Labs. 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.
  3. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. The United States employs an estimated 48% of people shaping AIWe estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.

Sources and credits

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

  1. [1312.4400] Network In Network arxiv.org. Checked 21 October 2026.
  2. Min Lin | Mila mila.quebec. Checked 21 October 2026.
  3. On Evaluating Adversarial Robustness of Large Vision-Language Models sail.sea.com. Checked 21 October 2026.
  4. NeurIPS 2024 Awards neurips.cc. Checked 21 October 2026.
  5. NUS Presidential Young Professor Kenji Kawaguchi wins Best Paper Award at NeurIPS 2024 - NUS Computing comp.nus.edu.sg. Checked 21 October 2026.
  6. There May Not be Aha Moment in R1-Zero-like Training — A Pilot Study sail.sea.com. Checked 21 October 2026.
  7. Beyond backpropagation: JAX's symbolic power unlocks new frontiers in scientific computing developers.googleblog.com. Checked 21 October 2026.
  8. World Models vs. Million LoRAs: 2 AI Labs, 2 Bets on Continual Learning fpfounderpark.substack.com. Checked 21 October 2026.