PeopleofSuperintelligence

Researcher

Eric Elmoznino

Eric Elmoznino is an AI researcher whose work connects visual representations, language-model inference, and compositionality. He studied computer engineering at the University of Toronto, developed computer-vision models at ModiFace, and researched human vision at Johns Hopkins before coauthoring an ICLR 2024 honorable-mention paper and a 2025 theory of representational compositionality. He is a computer-science PhD student at Mila and a part-time AI Student Researcher at Google's Paradigms of Intelligence team.

Now at
Google
Badges

Timeline

  1. 2012–2014

    Corporal in the Canadian Armed Forces

    He serves as a corporal in the Infantry Division of the Canadian Armed Forces Reserve from 2012 to 2014.

  2. 2012–2019

    Health sciences to computer vision

    He studies health sciences at Dawson College from 2012 to 2014, then earns a BASc in computer engineering at the University of Toronto in 2019. His capstone team builds an app that transcribes nearby speech and makes conversations searchable; from 2017 to 2019, he develops augmented-reality computer-vision models at ModiFace.

  3. 2019–2022

    Cognitive science at Johns Hopkins

    He completes an MA in cognitive science at Johns Hopkins in 2020 and works as a research assistant in its Cognitive Science Department from 2020 to 2022. With advisor Mick Bonner, he studies human vision.

  4. 2024

    High-dimensional models of vision

    In 2024, he and Michael F. Bonner report that neural networks with high-dimensional image subspaces show better generalization when predicting cortical responses to held-out stimuli in monkey electrophysiology and human fMRI data.

  5. 2024

    An ICLR honorable mention

    With six coauthors, he publishes Amortizing intractable inference in large language models at ICLR 2024. The work uses amortized Bayesian inference and generative flow networks to sample from intractable posteriors, and it receives an ICLR 2024 Honorable Mention.

  6. 2025

    Representational compositionality at ICML

    At ICML 2025, he coauthors Towards a Formal Theory of Representational Compositionality, a quantitative definition grounded in algorithmic information theory that treats a representation as expressive and describable as a simple function of parts.

  7. 2025–2026

    A visiting role at IBS

    In 2025 and 2026, he is a visiting researcher at the Institute for Basic Science's Center for Neuroscience Imaging Research under Hakwan Lau, working on brain-computer interfaces with fMRI and large language models. His 2026 paper Can In-Context Learning Support Intrinsic Curiosity? tests whether sequence models can use in-context learning as update-free world models for curiosity-driven data collection. He gives a seminar titled “Compression & Intelligence” at the center in July 2026.

  8. Since 2024

    Research at Google and Mila

    He continues a computer-science PhD at Mila under Yoshua Bengio and Guillaume Lajoie, a program he begins in 2022. He is a part-time AI Student Researcher on Google's Paradigms of Intelligence team. He joins the team in 2024 and studies open-ended knowledge acquisition with large sequence models capable of in-context learning.

Career

  1. Since 2024GoogleAI Student Researcher
  2. 2025–2026Institute for Basic ScienceVisiting Researcher
  3. 2020–2022Johns Hopkins UniversityResearch Assistant
  4. 2017–2019ModiFace
  5. 2012–2014Government of CanadaCorporal

Education

  1. From 2022MilaPhD
  2. 2020Johns Hopkins UniversityMA in cognitive science
  3. 2019University of TorontoBASc in computer engineering
  4. 2014Dawson College

From the biography.

  1. 23% of people shaping AI who left a university for a company later returned to a universityIncludes Eric Elmoznino. We estimate that 468 of the 2,011 people shaping AI who left a university for a company (23%) later returned to a university after a median of 5 years, and 163 of 829 (20%) made the reverse round trip.
  2. 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.
  3. 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.
  4. 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.

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 Eric Elmoznino yet. Send us a free-licensed photo.

  1. Eric Elmoznino ericelmoznino.github.io. Checked 21 October 2026.
  2. High-performing neural network models of visual cortex benefit from high latent dimensionality | PLOS Computational Biology journals.plos.org. Checked 21 October 2026.
  3. Amortizing intractable inference in large language models proceedings.iclr.cc. Checked 21 October 2026.
  4. ICLR 2024 Outstanding Paper Awards blog.iclr.cc. Checked 21 October 2026.
  5. Towards a Formal Theory of Representational Compositionality proceedings.mlr.press. Checked 21 October 2026.
  6. Seminar | IBS Center for Neuroscience Imaging Research cnir.ibs.re.kr. Checked 21 October 2026.
  7. Eric Elmoznino | Mila mila.quebec. Checked 21 October 2026.