Meta Core Interpretability Researcher Sonia Joseph Leaves to Launch NeoLab

soniajoseph_ · x · 2026-07-30

Sonia Joseph announced she has left Meta to build a frontier neolab focused on the physical world, interpretable foundation models, and scientific simulation. During her two-year tenure at Meta, she led the company's internal interpretability community and conducted research on the JEPA team. She is currently assembling a world-class founding team from leading AI and academic labs.

Related event: Ex-Meta Researcher Sonia Joseph Found World Mechanics(5 posts)→

Original post →

More from Companies & People

Companies & People channel →