Defending Against Decision Degradation with Full Spectrum Model Monitoring - Case Study and AMA

Explore a comprehensive case study and AMA session on defending against decision degradation through full-spectrum model monitoring at Lyft. Delve into the critical importance of preventing model degradation in high-stakes decision-making processes, including real-time pricing, physical safety classification, and fraud detection. Learn about Lyft's investment in building a robust model monitoring solution over the past two years. Discover their suite of approaches, including real-time feature validation, performance drift detection, anomaly detection, and model score monitoring. Gain insights into the cultural changes required to encourage ML practitioners to effectively monitor their models. Examine the tangible impact of Lyft's monitoring system in catching and preventing problems. Led by Mihir Mathur, Machine Learning Product Lead at Lyft, this 56-minute talk offers valuable lessons from a leading expert in the field of machine learning and AI-powered decision-making.

Defending Against Decision Degradation with Full Spectrum Model Monitoring Case Study and AMA