Ray + Arize - Why you Need Model Observability
summary
Demo shows a Lending Club loan-initiation fraud classifier served with Ray Serve and logged to Arize, capturing inputs, predictions and delayed actuals with under ten lines of code. Investigation of a California slice reveals many false positives tied to low loan amounts and feature shifts like purpose and loan_amount versus training, leading to a retrain. Platform adds distribution checks, real-time alerts, slice-level explainability and customers catch issues within 15 days.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 0:46The ML Infraspace
- 2:29Ray Arize Example
- 3:41Model Description
- 4:34Arize Platform
- 9:43Troubleshooting
- 10:58Why you need observability
related talks
29mLeveraging the Possibilities of Ray Serve
22mRay, a Unified Distributed Framework for the Modern AI Stack
18mAnti-Money Laundering ML Modelling Approach
42mRay: A distributed system for emerging AI applications
48mResponsible AI
13mRay and Anyscale: An Optimization Journey
06m[Flash talk] Fraud detection at scale using AI @ Razorpay
21mRay Train on Anyscale: Observability and Reliability at Scale