Alexandros G. Dimakis

Alexandros G. Dimakis is a Professor in the Department of Electrical Engineering and Computer Sciences at the University of California, Berkeley, with a joint appointment at the National AI Institute for Foundations of Machine Learning. He co-founded BespokeLabs.ai, focusing on data-centric AI solutions. His research spans Generative AI, Information Theory, and Machine Learning, with applications in inverse problems, compressed sensing, and large-scale data systems.

Dimakis's work explores fundamental machine learning paradigms including diffusion models, adversarial networks, and neural network verification. His recent publications demonstrate innovative approaches to pretraining data detection, multimodal dataset curation, and solving ill-posed inverse problems using generative priors. This research has significant implications for medical imaging, computational biology, and natural language processing.

Key scientific contributions have earned him recognition as an IEEE Fellow (2022) and recipient of the James Massey Award (2018). He received the Faculty of the Year award in 2022 from the MS Program in Information Technology Management. His research appears in premier venues including NeurIPS, ICLR, Nature Communications, and CVPR.

Dimakis actively mentors graduate students, leading a research group of PhD candidates exploring generative models, theoretical machine learning, and scientific applications. Current projects include OpenThoughts (reasoning dataset curation) and DeepSeek-R1 analysis. He serves on technical program committees for AAAI, NeurIPS, and ICML, and chaired MLSys 2021.