BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CAIL//Events//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:CAIL Events
BEGIN:VEVENT
UID:2026-06-15-vivan-amin@cail.columbia.edu
DTSTAMP:20260823T060715Z
DTSTART:20260615T173000Z
DTEND:20260615T183000Z
SUMMARY:Vision Seminar: Vivan Amin - From Labs to Last Mile: Bridging Phys
 ical AI Research and Real-World Deployment
LOCATION:CSB 480
DESCRIPTION:Physical AI is moving beyond pure exploration and into early e
 xperimentation\, but a significant reliability gap still stands between re
 search progress and general-purpose deployment. This talk examines the fro
 ntier challenges in embodied intelligence\, from vision-language-action ar
 chitectures such as CogAct\, which separate cognition from action generati
 on to compute-efficient training approaches such as FLOWER\, as well as em
 erging directions in VLA+ (Rho-Alpha)\, world action models (PIDM)\, Vitra
  and video action models (Mimic) directions emerging from Microsoft Resear
 ch.\n\nThe session will focus on why deployment now requires an end-to-end
  Physical AI toolchain: simulation and synthetic data\, edge data capture\
 , scalable training\, evaluation\, and repeatable deployment workflows. It
  will also highlight Microsoft’s recently introduced Azure Physical AI T
 oolchain\, which connects physical assets\, simulation\, and cloud trainin
 g into enterprise-grade pipelines designed to support the full simulate-tr
 ain-evaluate-deploy lifecycle The talk offers a grounded view of where the
  field stands today\, where the hardest open questions remain\, and where 
 academia-industry collaboration can have the greatest impact on reliable P
 hysical AI deployment.\n\nhttps://cail.columbia.edu/events/2026-06-15-viva
 n-amin
URL:https://cail.columbia.edu/events/2026-06-15-vivan-amin
END:VEVENT
END:VCALENDAR