PeopleofSuperintelligence

Researcher

Adriana Romero Soriano

Adriana Romero Soriano is a computer vision researcher whose work spans graph learning, medical imaging, image generation and fairness in visual systems. Her PhD thesis at the University of Barcelona, under Carlo Gatta, focused on training deep neural networks for computer vision and appeared in 2015; she later spent two years as a postdoctoral researcher at Mila with Yoshua Bengio. She coauthored Graph Attention Networks at ICLR in 2018, Inverse Cooking in 2019 and a reinforcement-learning MRI sampling method in 2020. In 2021, she coauthored IC-GAN; in 2022, fairness indicators; in 2023, she became a Canada CIFAR AI Chair; in 2025, she coauthored reward-guided decoding for multimodal LLMs. As of October 2026, she is a research scientist at Meta FAIR, an adjunct professor at McGill and a core industry member of Mila.

Links
Google Scholar, X and GitHub

Timeline

  1. 2015

    A PhD and a Mila postdoc

    She earns a PhD at the University of Barcelona under Carlo Gatta; her thesis on training deep neural networks for computer vision appears in 2015. She then spends two years as a postdoctoral researcher at Mila with Yoshua Bengio.

  2. 2018

    Graph Attention Networks

    In 2018, she coauthors the Graph Attention Networks paper at ICLR.

  3. 2019

    Inverse Cooking

    In 2019, she coauthors Inverse Cooking, a system that generates recipes from food images.

  4. 2020

    Reinforcement learning for MRI

    In 2020, she coauthors a reinforcement-learning method for active MR k-space sampling.

  5. 2021–2022

    IC-GAN and fairness indicators

    In 2021, she coauthors Instance-Conditioned GAN (IC-GAN), which models complex image datasets through overlapping neighborhoods around each data point. In 2022, she coauthors three fairness indicators that quantify harms and biases in visual systems.

  6. 2023

    A Canada CIFAR AI Chair

    In April 2023, she becomes a Canada CIFAR AI Chair while serving as a core industry member of Mila and an adjunct professor at McGill.

  7. 2025

    Reward-guided multimodal decoding

    In 2025, she coauthors a reward-guided decoding method for multimodal LLMs, using reward models to improve visual grounding and control object precision and recall.

  8. As of October 2026

    Research scientist at Meta FAIR

    She is a research scientist in Meta's Fundamental AI Research team, an adjunct professor at McGill and a core industry member of Mila.

Career

  1. Since 2015MilaPostdoctoral Researcher
  2. NowMcGill UniversityAdjunct Professor
  3. NowFAIR (Meta)Research Scientist

Education

  1. 2015University of BarcelonaPhD

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

Sources and credits

Photo: We have no published photo of Adriana Romero Soriano yet. Send us a free-licensed photo.

  1. Adriana Romero Soriano - Publications sites.google.com. Checked 20 October 2026.
  2. Adriana Romero Soriano | Mila mila.quebec. Checked 20 October 2026.
  3. Instance-Conditioned GAN arxiv.org. Checked 20 October 2026.
  4. Fairness Indicators for Systematic Assessments of Visual Feature Extractors arxiv.org. Checked 20 October 2026.
  5. The Canadian Institute for Advanced Research (CIFAR) and Mila appoint three new Canada CIFAR AI Chairs mila.quebec. Checked 20 October 2026.
  6. Controlling Multimodal LLMs via Reward-guided Decoding arxiv.org. Checked 20 October 2026.