Case Study: Capital One cuts model deployment time from months to minutes with Seldon

How Capital One reduced model deployment time from months to minutes

Capital One, a leading US retail bank, faced month-long lag times in its ML deployment pipeline as Data Science teams depended heavily on Engineering to test, deploy or upgrade models, often requiring full application redeploys and extra developer resources. To address this, Capital One deployed Seldon Core to provide a robust, scalable model serving platform across multiple business areas.

Using Seldon as the underlying infrastructure, Capital One built a “Model as a Service” (MaaS) that containerised polyglot models, exposed them as REST/gRPC endpoints, and integrated CI/CD, auditing, logging and monitoring to meet governance needs. Seldon enabled zero-downtime refreshes and rapid, self-service deployments, cutting model deployment/update time from months to minutes and delivering value across fraud, marketing, finance and customer service while improving explainability and compliance.

Capital One

Steve Evangelista

Director of Product Management