Инфраструктура и платформы данных в эпоху ИИ-агентов: как масштабироваться

   • The Infrastructure Nightmare Nobody Is Tal...   This text is an interview with Emma, ​​Head of Data Platform Infrastructure at OpenAI, who discusses the impact of rapid AI advancements on the company's internal processes. The focus is on the use of autonomous agents and Codex models to automate routine tasks such as software releases, bug fixes, and user support. Emma highlights the growing gap between the speed of application development and the capabilities of infrastructure teams, who are forced to implement safeguards against unintentionally destructive code generated by AI. To address these issues, she proposes a shift to a multi-agent architecture, where specialized agents independently verify and operate systems. The author calls on IT industry leaders to invest in their own testing libraries (evals) and adapt workflows to the new realities of scaling. Ultimately, the source describes a future in which collaboration between teams becomes entirely autonomous and agent-based, requiring engineers to take new approaches to managing complexity.