Ready cloud environment for your application: running K8s + VM + Database in two hours
In two hours you will build a working real-time fraud detection system for card payments on your own cloud. The legacy card authorization switch stays on a virtual machine, just as it does in every bank. Around it you will deploy Kafka, Redis and PostgreSQL from the catalog, and run a scoring service in your own Kubernetes cluster that decides on every transaction in a fraction of a second. A small language model running inside your cloud explains each blocked transaction in plain English, and the card data never leaves your perimeter. Then you will launch a fraud attack and watch the system stop it.
The second half is about running all of this in production: a Black Friday traffic spike, the primary database node going down, and the authorization switch moving to another server in the middle of the payment flow. Everything runs on Cozystack, an open-source CNCF platform that turns ordinary servers into a private cloud. All you need is a laptop with a browser and kubectl. Your environment stays with you for another month, together with a support chat and a Cozystack certification assignment.
Speaker: Timur Tukaev has run Linux on his desktop since 2007. He is an open-source enthusiast and a maintainer and community manager of Cozystack, a CNCF Incubating project. He co-founded Ænix. He likes experimenting with Kubernetes and finding unusual ways to rebuild it.