About
Sergio Cecarini
15+ years in DevOps and Platform Engineering. I have built and operated infrastructure at scale - from bare metal in co-los to Kubernetes clusters serving production AI workloads.
I started StackPulse because the conversation in AI infrastructure is dominated by vendors and hype. I wanted a place for practitioners to share what actually works, what actually breaks, and what actually matters when you are running AI in production.
Everything I write comes from hands-on experience. When I write about vLLM monitoring, I have instrumented production vLLM servers. When I write about FinOps, I have chased down cloud bills with seven-figure surprises. No thoughtpieces. No vendor briefings. No fluff.
What This Publication Covers
Monitoring, evaluation, hallucination detection, RAG observability, vLLM, and the infrastructure that keeps production AI running.
Cloud cost optimization, GPU spend management, context window efficiency, reserved capacity, and cost-aware architecture.
Kubernetes, Prometheus, Grafana, eBPF, and the tools and patterns that separate production-ready AI from science projects.