I build the platform layer that turns AI from a demo into a system teams can depend on — model serving, retrieval pipelines, agent and MCP tooling, and the observability and infrastructure-as-code that keep it reliable. After 15+ years across cloud infrastructure, DevOps and SRE, I now lead AI platform engineering: designing agent-driven platforms on Kubernetes with LLMs, and making the engineers around me faster and safer in the process.

My work lives at the intersection of infrastructure and machine learning — how AI systems are actually built, deployed, and operated in production, especially inside regulated and data-sovereign environments. I'm AI-native in how I work: coding agents like Claude Code are part of my daily workflow, and I build the tooling — MCP servers, hardened IaC, eval harnesses — that makes them effective.

Outside of work, I'm a part-time trader. Less interested in prediction, more in understanding market behavior, risk, and the psychology behind decision-making. For me, trading is another form of systems thinking — observing patterns, constraints, and failures over time.

I like tinkering: trying tools, writing small scripts, automating things that probably don't need automation, and occasionally breaking systems just to understand how they fail. This site is where I document ideas, experiments, and things I learn along the way.