Former Meta researcher launches a stealth physical-AI lab focused on interpretability
soniajoseph_ · x · 2026-07-29
A former Meta researcher says she has left to build a stealth “frontier neolab” for the physical world.
- The new company focuses on interpretable foundation models and scientific simulation for physical AI.
- She says the team is recruiting top researchers and engineers, with a bar that includes multiple first-author conference papers for researchers and tier-1 startup experience for engineers.
- The thesis is that interpretability should be part of the full lifecycle of physical AI systems: training, verification, and trust.
- The company is structured as a Public Benefit Corporation and says it wants to support open research and a broader ecosystem around physical-world data and safety.
- Target applications mentioned include climate science, astrophysics, supply chains, power grids, and robotics.
The post is both a talent-update signal and a snapshot of where “physical AI” startups are positioning themselves.
More from Companies & People
- The new CTO job: agent management and QA loops, 'cable management' instead of coding — HankYeomans · 2026-10-16
- Reddit debate: Are AI agent startups making real revenue or just burning VC money? — Pure_Armadillo4339 · 2026-10-16
- Paul Graham on investor rejections: dumb reasons, dislike — or both — dbasch · 2026-10-16
- Big Tech Is Killing Agentic AI — and Enterprises Are Letting It Happen — DavidLinthicum · 2026-10-16
- Pivotal opens Q1 2027 AI Safety Research Fellowship: 15 weeks in London, £15-18k stipend — peterbhase · 2026-10-16
- Iason Gabriel named Principal Scientist and Director of AGI & Society at DeepMind Institute — IasonGabriel · 2026-10-16