$ cat /etc/operating-principles.md
In modern infrastructure, emitting logs is easy. Emitting actionable signals is hard. I build systems that prioritize highly contextual, low-volume alerting. If a metric doesn't directly impact the user or the business, it doesn't belong on a dashboard. Complexity is a liability; clarity is a feature.
Generative AI and Agentic systems are ultimately distributed compute problems. They have context limits, latency spikes, token economics, and trust boundaries. Treating AI as an API call leads to fragile systems. Treating AI as infrastructure leads to resilient, observable, and scalable platforms.
Claims require proof. A resume is a marketing document; a production deployment is evidence. I value reproducible infrastructure (IaC), zero-drift configuration, and open community participation over theoretical knowledge. If it's not in source control, it doesn't exist.
Manual intervention is a bug. Platform engineering exists to pave paths and remove friction. If an engineer has to open a ticket to provision resources, the platform has failed. We optimize for self-service, autonomous remediation, and zero-touch deployments.
~ vishal-gunjal