About
Berivan Isik is a prominent research scientist at Google DeepMind, specializing in generative AI with a focus on efficient training and finetuning of large language models, pretraining data valuation, scaling laws, differential privacy, and machine unlearning to advance trustworthy AI systems. She earned her PhD in Electrical Engineering from Stanford University in 2024, where she was co-advised by professors Sanmi Koyejo and Tsachy Weissman, and her groundbreaking work earned her prestigious awards including the Stanford Graduate Fellowship and the Google PhD Fellowship in Machine Learning. With a prolific publication record in top venues like ICLR, AISTATS, and ICML—often featuring oral presentations and collaborations on high-impact topics such as federated learning, model compression, and information-theoretic approaches to AI—Isik has established herself as a leading voice in scalable and responsible machine learning, contributing directly to cutting-edge developments in models like Gemini while advocating for privacy-preserving techniques in the era of massive-scale AI.