Understanding Kubernetes Architecture with Mohit Dharmadhikari

Kubernetes Architecture Masterclass

From Chaos to Kubernetes — A First-Principles Journey into How Kubernetes Works

Kubernetes can feel overwhelming when learned as a collection of components, YAML files, and commands.

This masterclass takes a different path.

Instead of starting with “What is a Pod?” or “How do I write a Deployment?”, we start with a more fundamental question:

What problems must a distributed system solve when thousands of workloads are constantly changing, failing, moving, and recovering?

From there, we progressively derive the architectural ideas that lead to Kubernetes.

We explore why a system needs a shared API boundary, durable state, controllers, reconciliation loops, scheduling, node agents, service discovery, and mechanisms for continuously converging the real world toward a desired state.

Through interactive simulations and architectural visualizations, we follow the journey from a simple workload to a distributed system capable of managing constantly changing infrastructure.

What you'll explore

• Why Kubernetes needs a control plane
• How the API Server and etcd form the foundation for cluster state
• Why Kubernetes uses declarative desired state
• How controllers and reconciliation loops continuously repair drift
• How the scheduler decides where workloads should run
• What actually happens when a node or workload fails
• How the kubelet keeps workloads running on individual nodes
• Why Pods need stable networking identities
• How Services and EndpointSlices connect clients to changing workloads
• How independent controllers cooperate without requiring one component to understand the entire cluster
• How these ideas come together to form the architecture we recognize as Kubernetes

The goal is not to memorize Kubernetes components.

The goal is to build a mental model strong enough that the components make sense.

By the end of the masterclass, Kubernetes should no longer look like a collection of mysterious boxes. You should be able to look at its architecture and understand why each major piece exists, what problem it solves, and how the pieces cooperate to maintain a distributed system.

First principles → mental models → real understanding.