Production Machine Learning Monitoring: Outliers, Drift, Explainers & Statistical Performance
summary
Covers production monitoring for ML using Kubernetes microservice patterns and standard observability techniques, plus advanced ML-specific monitors: concept-drift detectors, outlier detectors, adversarial detectors and AI explainers. Shows a hands-on example training an image-classification model, deploying it as a Kubernetes microservice, and integrating those monitoring components as architectural patterns and standardized interfaces to scale across many heterogeneous models.
AI-written summary · about
related talks
16mMLOps — Understanding and Fighting Model Degradation in Production
47mMachine Learning in Production: An Intro to MLOps.
31mMonitoring machine learning models in production
39mAn SRE Approach to Monitoring ML in Production
29mDemystifying Machine Learning in Production: Reasoning about a Large-Scale ML Platform
18mReliable Maintenance of Machine Learning Models
44mDevOps for Machine Learning - why is it different?
26mCloud Native Machine Learning Systems at Day Two and Beyond