Herbert Woisetschläger

Herbert is a research scientist with the IBM Research Zurich Lab, where he is part of the AI Foundations department. His work spans efficient methods for training and inference of LLMs as well as memory, routing, and tooling for AI agents. The common thread of his work is making deep learning systems more capable with fewer resources and overall more economical to run.

Before joining IBM Research, Herbert completed a PhD in computer science at the Technical University of Munich (TUM), specializing in distributed and cost-efficient machine learning. His work has been published at leading machine learning and computer systems conferences (e.g., NeurIPS, ICML, ICLR, ACM Middleware). During his doctoral studies, he did internships with the Granite Data & Tools group at the IBM T.J. Watson Research Center in Yorktown Heights.

Herbert came to research from industry. He spent two years as a management consultant with Capgemini Invent, focusing on performance and cost optimization programs for consumer goods retailers, financial institutions, and manufacturers. He also worked at Detecon, Inc. in Silicon Valley on corporate startup incubation and accelerator programs.