Streaming to Scale - Real-Time Infrastructure for AI

Learn how to transform streaming infrastructure from basic durable queues into an intelligent backbone that powers both production AI workloads and research at scale in this conference talk by Aravind Suresh from OpenAI. Discover the critical role of real-time streaming pipelines in modern AI systems that demand fast, reliable, and adaptable infrastructure for continuous model updates. Explore practical strategies for scaling Kafka and Flink to handle high-throughput AI workloads while supporting complex AI products like ChatGPT and Sora. Master the implementation of abstraction, automation, and observability techniques to maintain system resilience during rapid growth phases, including insights from managing 5× quarterly expansion. Gain actionable insights into building streaming infrastructure that serves as the nervous system of modern AI platforms, enabling real-time intelligence and rapid innovation for AI systems, data infrastructure, and research pipelines.