Model as a Service for Real-Time Decisioning
summary
Capital One builds a Model-as-a-Service stack that lets data scientists deploy models without engineering handoffs, using Seldon Core on Kubernetes to serve H2O/Scikit-learn/TensorFlow artifacts via REST or gRPC. A model deployment service runs CI/CD, static security checks, containerization, automated blue/green tests and Slack/email notifications; a routing layer supports A/B, champion–challenger and multi-armed bandits and a features as a service cache for sub-second lookups. Goal: shrink deployments that used to take weeks or months into self-service, iterative releases.
AI-written summary · about
chapters (9)
- 0:00Introduction to the platform
- 0:44Addressing the need for scale
- 2:36Model as a Service vision
- 5:54Architecture challenges
- 8:00Technical architecture details
- 12:15Data scientist workflow
- 15:10Serving, monitoring, and APIs
- 18:25Benefits and summary
- 19:29Question and answer session
related talks
46mProduction Model Deployment
34mServing ML Models at Scale with Seldon and Kubeflow
27mDeploying distributed AI and machine learning in financial services
04mInterview
27mDeveloping and Deploying Models at Scale — without Spark!
44mBuilding Multi-Tenant Compute Systems in the Enterprise
49mImpact on business and society
32mAutomating Production Production Level Machine Learning on AWS