Javier Abad

about

I'm a Research Scientist at Google DeepMind working on Gemini post-training. I completed my PhD in machine learning as a Fellow at the ETH AI Center advised by Fanny Yang.

My PhD research focused on language model alignment and safety, privacy-preserving machine learning, and trustworthy machine learning more broadly. During that time, I also worked on pretraining visual generation models at Apple and on language model distillation at Google DeepMind. Before my PhD, I was a Research Scientist at Featurespace. Earlier, I spent time at the University of Cambridge, researching conformal prediction with Adrian Weller MBE and causal inference with Mihaela van der Schaar.

research

Full list on Google Scholar.

^* equal contribution

• Efficient randomized experiments using foundation models
  Piersilvio De Bartolomeis, Javier Abad, Guanbo Wang, Konstantin Donhauser, Raymond M. Duch, Fanny Yang, and Issa J. Dahabreh.
  Conference on Neural Information Processing Systems (NeurIPS) · 2025 · Paper · Code
• Copyright-protected language generation via adaptive model fusion
  Javier Abad, Konstantin Donhauser, Francesco Pinto, and Fanny Yang.
  International Conference on Learning Representations (ICLR) · 2025 · oral · Paper · Code · Talk
• Detecting critical treatment effect bias in small subgroups
  Piersilvio De Bartolomeis, Javier Abad, Konstantin Donhauser, and Fanny Yang.
  Conference on Uncertainty in Artificial Intelligence (UAI) · 2024 · Paper · Code
• Privacy-preserving data release leveraging optimal transport and particle gradient descent
  Konstantin Donhauser^*, Javier Abad^*, Neha Hulkund, and Fanny Yang.
  International Conference on Machine Learning (ICML) · 2024 · Paper · Code
• Hidden yet quantifiable: a lower bound for confounding strength using randomized trials
  Piersilvio De Bartolomeis^*, Javier Abad^*, Konstantin Donhauser, and Fanny Yang.
  International Conference on Artificial Intelligence and Statistics (AISTATS) · 2024 · Paper · Code
• Approximating full conformal prediction at scale via influence functions
  Javier Abad, Umang Bhatt, Adrian Weller, and Giovanni Cherubin.
  AAAI Conference on Artificial Intelligence · 2023 · oral · Paper · Code

news

• Jul 2026 Rejoined Google DeepMind as a Research Scientist working on Gemini post-training.
• Sep 2025 Joined Google DeepMind as a PhD Student Researcher working with the Gemma post-training team.
• Apr 2025 Presented Copyright-Protected Language Generation via Adaptive Model Fusion as an oral at ICLR 2025 in Singapore.
• Mar 2025 Started working at Apple on image and video generation for Apple Intelligence.
• Oct 2024 Visiting the Simons Institute at UC Berkeley.
• Feb 2023 Presented Approximating full conformal prediction at scale via influence functions as an oral at AAAI 2023 in Washington, DC.
• Dec 2022 Started my PhD at the ETH AI Center.