How Data Scientists Deploy Model Monitoring at Scale
from The Data Science Podcast with Fexingo: Analytics, Machine Learning, and Data-Driven Conversations Hosted by Fexingo
Episode 126 of The Data Science Podcast dives deep into model monitoring. Lucas and Luna explore why production ML systems degrade silently and how teams catch concept drift before it costs millions. They break down a real case: a mid-size fintech lender whose fraud model's false-positive rate crept up by 12 percent over three months without anyone noticing — until a monitoring alert caught it. The hosts walk through the monitoring stack: data quality checks, prediction distribution tracking, feature attribution drift, and alerting thresholds. They discuss tradeoffs between real-time and batch monitoring, open-source tools like Evidently AI and WhyLabs versus managed platforms, and how to build a monitoring culture that treats models like services. If you're a data scientist shipping models to production, this episode gives you a concrete framework for knowing when your model is lying to you. #ModelMonitoring #MachineLearning #DataScience #MLOps #ConceptDrift #ProductionML #DataQuality #FeatureDrift #EvidentlyAI #WhyLabs #FraudDetection #Fintech #ModelObservability #PredictionDistribution #Alerting #Technology #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo
Embed this episode
No transcript yet.