In 2020, Apress published my book Advanced Platform Development with Kubernetes, about five hundred pages that assembled a complete data platform out of open source on Kubernetes. The ideas held up. The specific technology did not. Weave Net is gone, ingress-nginx is retired, the Docker shim was removed from the kubelet, Kafka no longer needs ZooKeeper, and Elasticsearch changed its license in 2021. This series rebuilds the entire platform on what I would actually deploy now.

Every note is a complete build rather than a survey: the manifests, each configuration option that matters and the reason it is set that way, and what the component replaced. Read in order they assemble one platform, starting from generic VMs and running up through storage, ingress, databases, streaming, the lakehouse, observability, on-prem GPUs, and an MCP endpoint that gives agents grounded access to all of it. Any one of them also works on its own if you only need that piece.

The other thing that changed is who does the operating. The reason most teams rent a data warehouse was never that building one is hard. It is that running one is a full-time job, and that is the part agents now take on. The opening note makes that case, and it runs underneath everything that follows.