Architect-to-Builder: a reliable way to ship with agents
Architect-to-Builder splits AI agent planning and implementation across two roles, joined by a living plan file, so the Builder codes before context is half gone.
Arv
I’m an engineer working on AI agents and the systems around them. At Playtronix, I lead engineering and build agent workflows that connect models to our documentation, code, and tools.
I bring a decade of experience in distributed systems, high-load backends, and data infrastructure, including work at Freshworks and in crypto and fintech. My focus now is making LLM applications useful and reliable in practice. I write here about agents and what I learn building them.
Interests. AI agents, retrieval-augmented generation, and reliable LLM applications.
Writing
RSS feedService Evolution: From Monolith to Microservices
Why you should start with a monolith, avoid premature distribution, and introduce microservices only when complexity has a clear reason.
Kubernetes Cheat Sheet: kubectl, Deployments, and Services
Practical Kubernetes cheat sheet covering core concepts, Minikube, deployments, services, and common kubectl workflows.
Go for Experienced Programmers: Arrays, Slices, Maps, and Structs
A concise introduction to Go for experienced programmers: arrays, slices, maps, structs, and the core mechanics that make the language small and fast to learn.
Reading
I keep a reading log of papers, talks, and posts I return to, on AI, systems, and building software.