Learning from Extremes: What Fraud-Fighting at Scale Can Teach Us About MLOps Across Domains

Explore the cutting-edge MLOps practices developed for large-scale anti-fraud platforms and their potential applications across various domains in this 29-minute conference talk. Discover how the extreme demands of fraud detection, including low-latency inference, feature freshness, and agile redeployment, have led to pioneering advancements in MLOps. Challenge the assumption that these intense problem-solving approaches are excessive for less operationally-demanding fields. Examine how a real-time-first approach simplifies architectures by eliminating complex pipelines. Gain insights into observability and replay technologies designed for rapid response to unpredictable attacks, and learn how these tools can enhance agility for ML teams across different sectors. Join Greg Kuhlmann, CEO of Sumatra, as he shares valuable lessons from fraud-fighting at scale and their broader implications for MLOps practices.