Writing

The world produces more noise than any of us can process. How I decide what actually deserves my attention, and how I built a feed that serves me instead of the other way round.

What Stoicism actually is, where it comes from, why an engineer should care, and where to start reading. Featuring a shipwrecked merchant from Kition.

The mindset that kept me sane through two years of running a consultancy, and why I think the word 'failure' is doing real damage to would-be founders.

Where LLM agents genuinely help in ops work today, the permission model we use before any agent gets near client infrastructure, and two failure stories that shaped it.

Portability is the reason everyone gives for multi-cloud and almost never the real one. What it actually costs, when it's genuinely justified, and what 'portable enough' looks like.

What eBPF actually is, why Cilium replaced kube-proxy and a mountain of iptables rules, what Hubble gives you for free, and when a boring VPC-native CNI is still the right answer.

How a four-person consultancy runs sustainable incident response: alert budgets, runbooks as the price of admission, honest SLAs, and severity levels that actually mean something.

Metrics, logs, and traces without the seven-figure vendor bill: how OpenTelemetry and a small Grafana stack get small teams surprisingly far.

Most Kubernetes overspend is requests nobody measured, nodes nobody shaped, and a bill nobody sees. All three are fixable.

An honest retrospective on Nepheli's first year: underpricing, wrong-fit clients, the sales pipeline nobody warned me about, and what I'd tell myself on day one.

What GitOps actually buys you, an honest ArgoCD vs Flux comparison, and how I pick between them per client.

What golden paths look like when you're four people, or consulting for twenty developers who just want to ship.

Static binaries, boring readability, and a concurrency model that fits ops work: how Go became the default language of cloud infrastructure, and when I still reach for Python or Bash.

Remote backends, locking, and why how you split your state files matters more than any module you'll ever write.

How I finally removed the long-lived AWS keys from my own site's deploy pipeline, and how OIDC federation actually works under the hood.

Exploring the emerging trend of vibe coding - a way of working that emphasizes flow states and intuition over strict planning in software development.

A comprehensive analysis of the differences between startup and corporate careers, helping you make an informed decision about your professional path.

An in-depth look at the current state of PC hardware, including GPU and CPU market updates, memory and storage trends, and actionable advice for builders.

A detailed walkthrough of my interview process, including preparation, execution, and follow-up.

Exploring Colima as a lightweight alternative to Docker Desktop for container development.

A comprehensive guide to IT certifications, their value, and how to approach them.

A detailed comparison of different Container Network Interface (CNI) solutions.