About me

I’m Rinor. I write here to share the things I have tried and the things that turned out to work well.

Lately, I’m mostly curious about the plumbing behind AI agents. At Solo.io, I build an event-driven agentic harness. Here I share how agents pass messages, manage context, stay stateless and keep humans in the loop, and how to make a harness more efficient with Code Mode, tool preloading and decision models. I’ve also been poking around the unsolved parts of the Model Context Protocol, especially authorization: how an agent acts on someone’s behalf with narrow permissions. On the side, I maintain MCP-Nest, a NestJS framework for MCP servers, and I tinker with retrieval, reranking, and low-latency voice interfaces.

Before agents took over my brain, my rabbit holes were networking and distributed systems: Kubernetes, Envoy, observability with OpenTelemetry, and service meshes. You’ll still find my long, hands-on series on those here. Those series are also how I came to write Istio in Action together with Christian Posta.

My old rule of thumb was “learn fast, apply, forget”, never using the same technology twice. I stick around with tools a bit longer now, but the habit of pulling complex systems apart hasn’t changed. Expect deep-dive tutorials, hand-drawn diagrams, and the occasional essay on staying in the “flow” zone while programming.

You can find my code on GitHub and my half-baked thoughts on X.