More Context Isn't Better - You Pay for the Extra Twice
Two AI projects taught me the same lesson in opposite ways. One ran three weeks and cost $10; the other took an hour and hit $100. Same model, same me - so what actually changed?
Thinking. Building. Learning.
A personal space for learning in public - through code, ideas, and imperfect progress.
Spark on Kubernetes - Running Spark as a Kubernetes-Managed Workload
This article shows how Spark behaves when spark-submit runs inside Kubernetes Deployments, and how this changes lifecycle control, driver behavior, and long-running job management.
He Was Just a Kid - A Story About Technical Debt
Technical debt is often described as a code problem. It's usually the result of decisions made under pressure. This story follows technical debt from its birth as a small compromise to its eventual repayment, exploring why it grows, why it lingers, and why it isn't always the villain we make it out to be.
Spark on Kubernetes: Understanding Driver Lifecycle in Client and Cluster Mode
A practical guide to understanding Spark driver lifecycle ownership on Kubernetes. Explore how client and cluster mode affect job execution, termination, and operational behavior through real examples.
Weeks Using AI in Backend Development: What Actually Helped
What actually happens when you start using AI in real backend development work? A few practical examples from a legacy system rewrite - including API analysis, test generation, and disposable tooling.