Stefano Soatto
Vice president and distinguished scientist in AWS Agentic AI; UCLA professor of computer science
- Vice president and distinguished scientist, AWS Agentic AI, Amazon Web Services (Amazon)13
- Professor of Computer Science (listed as on leave), University of California, Los Angeles (Samueli School of Engineering)23
Stefano Soatto is a vice president and distinguished scientist in the Amazon Web Services Agentic AI organization and a computer science professor at the UCLA Samueli School of Engineering, which lists Soatto as currently on leave. UCLA's Science Hub for Humanity and Artificial Intelligence lists Soatto as Vice President, Agentic AI, Amazon Web Services and Professor, UCLA Computer Science, and earlier UCLA pages describe Soatto as vice president of applied science for AWS AI. Soatto's recent papers with AWS colleagues cover memory layers for language models, such as Gated KalmaNet (CVPR 2026), and automatic design of multi-agent systems, such as EvoMAS (ICML 2026).12345
Work
Gated KalmaNet: A Fading Memory Layer Through Test-Time Ridge Regression (external site: arxiv.org)Published research
Co-author (last author) of the paper, listed on Soatto's Amazon Science page as a CVPR 2026 publication. The abstract describes a Kalman-filter-based layer that keeps more of the past than fading-memory models and says the code and 8B and 32B hybrid models are released under Apache 2.0.41
EvoMAS: Evolutionary Generation of Multi-Agent Systems (external site: arxiv.org)Published research
Co-author (last author) of the ICML 2026 paper, which evolves the configurations of LLM-based multi-agent systems through feedback-driven mutation and crossover.51
This profile covers public professional work only, from the sources below. It is not an endorsement or a ranking. To correct or remove it, use the correction form.