Enric Boix

Enric Boix is an Assistant Professor at the University of Pennsylvania, affiliated with The Wharton School and the Statistics and Data Science Department. His research focuses on the mathematical foundations of deep learning, including neural network training dynamics, inductive bias, adversarial robustness, and AI safety.

• Undergraduate: Princeton University (advised by Emmanuel Abbe)
• PhD: MIT EECS (advised by Guy Bresler and Philippe Rigollet)
• Postdoctoral: MIT Mathematics & Harvard CMSA

Current work investigates theoretical aspects of AI, such as:

• Feature learning in neural networks
• Model distillation and fine-grained expert architectures
• Chain-of-thought reasoning in LLMs
• Inductive bias in ResNets and Transformers
• NTK approximation validity and optimization dynamics
• Adversarial prompt vulnerabilities

His publications span top conferences like NeurIPS, ICLR, and EMNLP, emphasizing mathematical rigor in understanding deep learning phenomena. Awards include:

• NSF Graduate Research Fellowship
• Siebel Fellowship
• Apple AI/ML Fellowship

He has collaborated extensively with researchers including Mikhail Belkin, Emmanuel Abbe, and Philippe Rigollet, focusing on theoretical advancements to improve AI efficiency and trustworthiness.