Abstract: Modern Internet-scale services must identify and mitigate customer-impacting incidents quickly. Despite the development of many Root Cause Analysis (RCA) algorithms—including recent LLM-assisted solutions—existing approaches struggle with the “long tail” of novel failure modes and the sheer scale of telemetry. In this talk, I argue that the path forward requires a paradigm shift: from developing point-solution algorithms to a systems-first approach that enables rapid prototyping of diverse RCA techniques. I will introduce MoCE: a Mixture-of-Experts framework that provides high-level abstractions for failure diagnosis. MoCE lets developers express complex troubleshooting logic succinctly via its domain-specific operators while its underlying runtime system handles scalable telemetry processing. Finally, I will briefly describe how these abstractions enable reliable, autonomous agents for domain-specific data analytics tasks.
Bio: Vipul Harsh is a postdoctoral researcher at Conviva with Vyas Sekar and a visiting researcher at Carnegie Mellon University (CMU). His research interests lie at the intersection of networked systems, AI, and theory. His current research focuses on developing new abstractions and frameworks for building reliable and verifiable systems — from networked infrastructure to AI agents. He has also worked on datacenter topology, distributed monitoring, and parallel algorithms. His research has been published in top-tier CS conferences (SIGCOMM, NSDI, SPAA, among others); his work on Murphy was adopted into VMware’s network management suite and DRing/Starfish have directly influenced Amazon AWS’s datacenter architecture. He completed his Ph.D. at UIUC, where he worked with Brighten Godfrey, and holds an undergraduate degree from IIT Bombay. His thesis was nominated by UIUC for the ACM SIGCOMM dissertation award.
