Can AI Agents Ship Code Without Human Review? | Pavel Sviridov

Pavel Sviridov, CTO of Nebius Education, joins Singula Talks to discuss AI coding agents, pull request review, security risks, and the engineering decisions that still need human judgment. 00:00 — AI Agent Gone Rogue: A Scary Deployment Story 01:17 — Meet Pavel Sviridov: CTO at Nebius Education 03:09 — From Cybersecurity to Software Engineering 06:12 — How AI Is Changing Cybersecurity 08:24 — The Future of Programming: Do Engineers Still Need to Code? 12:31 — Can AI-Generated Code Go Straight to Production? 18:17 — What Skills Will Software Engineers Need in the AI Era? 23:13 — Hiring Engineers Who Work with AI Agents 27:40 — Building AI-Ready Architecture with Domain-Driven Design 32:04 — When AI Agents Make Dangerous Mistakes 37:30 — How a CTO Learns New Technologies with AI 44:15 — The Rise of Personalized Software 50:45 — Can Anyone Build Their Own Apps with AI? 53:00 — Replacing Developers with AI-Powered Workflows 56:18 — Why We Removed Coding Tests from Interviews 62:20 — The Junior Developer Problem: Who Will Become Tomorrow's Seniors? 66:30 — Will AI Reduce the Number of Software Engineers? 68:00 — Does Code Quality Still Matter in the AI Era? 72:10 — How AI Agents Changed the CTO's Daily Workflow 75:35 — Building a Second Brain with Claude and Obsidian 78:20 — Career Decisions, Leadership and Managing Complexity 82:40 — Systems Over Goals: Lessons for Tech Leaders 84:30 — Interview Red Flags and the Future of Engineering