The session#
The meetup was organized through a local tech community group in Accra. I was asked to do a 30-minute session on Kubernetes internals — specifically how the scheduler makes decisions and what happens when a node goes down.
I walked through the scheduling pipeline, showed some real YAML examples, and demoed a node failure scenario in a test cluster. The room was mixed — some students who had never touched Kubernetes, some engineers with production experience. That range made it more interesting to navigate.
What worked#
Leading with "here's what fails and why" instead of "here's what Kubernetes does" got more attention. Engineers respond to problems better than architectures. The failure scenario demo landed better than the theory slides.
The follow-through#
Three questions from the floor became the seeds for blog posts:
- How does the scheduler handle pod affinity and anti-affinity at scale?
- What's the right mental model for node vs. pod failure?
- When does a DaemonSet actually matter?
Two people reached out the week after — one asking about breaking into cloud engineering, one asking about the Scratchcode bootcamp. That's the part of community work that compounds slowly but adds up.
The real win
Teaching something forces you to understand it better. Preparing this session tightened my own mental model of Kubernetes scheduling considerably.