Angéline N. Pouget
Curious researcher. Sports enthusiast. Lifelong learner.

Welcome! I’m Angéline, a research engineer at Google DeepMind in Zurich, very curious about the world around me and deeply passionate about diving headfirst into fascinating topics. Currently, my focus is on data-centric AI research, particularly on improving multimodal understanding in large language models.
Previously, I was a Master’s student at ETH Zürich and a visiting graduate student at the University of Toronto, advised by Nicolas Papernot. I’ve also had the opportunity to intern across various industries including tech, finance and consulting, gaining diverse insights along the way.
Outside of my professional pursuits, I’m a certified personal trainer and a dedicated hybrid athlete, triathlete and BQ marathoner, passionate about pushing my limits through swimming, cycling, running and strength training.
news
Jul 30, 2025 | I passed the certification exam and I’m now a National Academy of Sports Medicine (NASM) Certified Personal Trainer. It has been a fun journey diving into papers and research in a completely different field and learning more about running faster, lifting heavier, and minimizing injury risk. |
---|---|
May 29, 2025 | My Master’s thesis paper on suitability filters for classifiers has been accepted as a spotlight poster to ICML 2025 and selected for an oral (code here). I’m very thankful for the great collaboration with Mohammad Yaghini, Stephan Rabanser and Nicolas Papernot. |
Jan 06, 2025 | I just re-joined Google DeepMind in Zürich to work on multimodal understanding in large language models. |
Sep 25, 2024 | Our paper on cultural diversity in contrastive vision-language models has been accepted to NeurIPS 2024. |
Jul 15, 2024 | Our paper on cultural diversity in contrastive vision-language models has been accepted with an oral presentation at the Data-centric Machine Learning Research workshop at ICML 2024. Looking forward to giving a brief talk, participating in a panel, and interesting poster session discussions. |