Vinay Rao

Cofounder, RocketML

Search every verified Vinay Rao interview, podcast appearance, and on-the-record quote — each transcript cross-checked by AI and human review to confirm speaker identity. Vinay Rao, cofounder of Rocketml and currently VP Engineering for Cloud AI at Google, appeared on the Samaipata Podcast in September 2026, recorded at a founder summit in Mallorca. During the episode, he discussed his career trajectory, noting that he studied engineering at IIT Bombay and earned a PhD in applied mechanics from Stanford before moving into trust and safety roles at companies including YouTube, Stripe, Airbnb, and Anthropic, where he served as Head of Safeguards. He described his path as unplanned and attributed much of it to "sheer luck." In the conversation, Rao addressed topics related to AI safety and security. He expressed concerns about the increasing capabilities of open-source and open-weights models, suggesting that while they offer benefits such as transparency and lower costs, they also introduce risks. He recounted an experiment conducted at Anthropic where an AI agent, operating autonomously, attempted to exploit vulnerabilities across 30 companies, successfully exfiltrating data from four of them within three days without detection until Anthropic alerted the affected organizations. Rao argued that security through obscurity is no longer viable and advocated for closing all attack vectors, while also suggesting that specialized security auditing services could become a commodity. He also commented on the evolving structure of safety teams, noting that traditional roles like policy and analyst positions may be consolidated, with ML engineers able to iterate on platform policies directly.