About

Hello! I'm a postdoctoral researcher at Princeton University, working with Professor Sanjeev Arora at Princeton Language and Intelligence (PLI). My current research is on the reasoning abilities of large language models, including mathematical problem solving, formal theorem proving, and post-training methods, along with the theory of how transformers generalize.

Until August 2025, I was a Senior Research Scientist at Google DeepMind, where I worked on tool use for Gemini. I completed my PhD in Mathematics at the University of Maryland, co-advised by Professor Tom Goldstein (CS) and Professor Wojtek Czaja (Math), where my research focused on robustness and security for deep learning models, spanning adversarial robustness, data poisoning, and privacy attacks against language and vision models in federated learning.

Publications

* denotes equal contribution; † denotes equal senior authorship.

• Escaping the Cognitive Well: Efficient Competition Math with Off-the-Shelf Models.
  Xingyu Dang, Rohit Agarwal, Rodrigo Porto, Anirudh Goyal, Liam Fowl*, Sanjeev Arora. AI for Math Workshop, ICML 2026; under review, NeurIPS 2026.
• Goedel-Architect: Streamlining Formal Theorem Proving with Blueprint Generation and Refinement.
  Jui-Hui Chung, Ziyang Cai, Zihao Li, Qishuo Yin, Rohit Agarwal, Simon Park, Rodrigo Porto, Narutatsu Ri, et al., Liam Fowl*†, Sanjeev Arora. AI for Math Workshop, ICML 2026.
• Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervision.
  Yinghui He, Simran Kaur, Adithya Bhaskar, Yongjin Yang, Jiarui Liu, Narutatsu Ri, Liam Fowl, Abhishek Panigrahi, Danqi Chen, Sanjeev Arora. ICML 2026 RLxF Workshop; Spotlight, ICML 2026 AI for Math Workshop.
• Rethinking On-Policy Self-Distillation for Thinking Models.
  Simran Kaur, Narutatsu Ri, Yinghui He, Liam Fowl, Sanjeev Arora. ICML 2026 AI for Math Workshop; under review, NeurIPS 2026.
• To See the Unseen: On the Generalization Ability of Transformers in Symbolic Reasoning.
  Nevena Lazic, Liam Fowl, András György, Csaba Szepesvári. COLM 2026.
• Measuring the Limits of Continual Learning for LLMs.
  Nimit Kalra, Narutatsu Ri, Zerzar Bukhari, Ang Li, Sanae Lotfi, Liam Fowl, Micah Goldblum. Workshop on Compositional Learning, ICML 2026.
• Decepticons: Corrupted Transformers Breach Privacy in Federated Learning for Language Models.
  Liam Fowl*, Jonas Geiping*, Steven Reich, Yuxin Wen, Wojtek Czaja, Micah Goldblum, Tom Goldstein. ICLR 2023.
• Sleeper Agent: Scalable Hidden Trigger Backdoors for Neural Networks Trained from Scratch.
  Hossein Souri*, Liam Fowl*, Rama Chellappa, Micah Goldblum, Tom Goldstein. NeurIPS 2022.
• Fishing for User Data in Large-Batch Federated Learning via Gradient Magnification.
  Yuxin Wen*, Jonas Geiping*, Liam Fowl*, Micah Goldblum, Tom Goldstein. ICML 2022.
• Can Neural Nets Learn the Same Model Twice? Investigating Reproducibility and Double Descent from the Decision Boundary Perspective.
  Gowthami Somepalli, Liam Fowl, Arpit Bansal, Ping Yeh-Chiang, Yehuda Dar, Richard Baraniuk, Micah Goldblum, Tom Goldstein. CVPR 2022 (Oral).
• Robbing the Fed: Directly Obtaining Private Data in Federated Learning with Modified Models.
  Liam Fowl*, Jonas Geiping*, Wojtek Czaja, Micah Goldblum, Tom Goldstein. ICLR 2022.
• Adversarial Examples Make Strong Poisons.
  Liam Fowl*, Micah Goldblum*, Ping-yeh Chiang*, Jonas Geiping, Wojtek Czaja, Tom Goldstein. NeurIPS 2021.
• Witches' Brew: Industrial Scale Data Poisoning via Gradient Matching.
  Jonas Geiping*, Liam Fowl*, Ronny Huang, Wojtek Czaja, Gavin Taylor, Tom Goldstein. ICLR 2021.
• Adversarially Robust Few-Shot Learning: A Meta-Learning Approach.
  Micah Goldblum*, Liam Fowl*, Tom Goldstein. NeurIPS 2020.
• MetaPoison: Practical General-purpose Clean-label Data Poisoning.
  Ronny Huang*, Jonas Geiping*, Liam Fowl, Gavin Taylor, Tom Goldstein. NeurIPS 2020.
• Adversarially Robust Distillation.
  Micah Goldblum*, Liam Fowl*, Soheil Feizi, Tom Goldstein. AAAI 2020.