Hyon Lee

I co-founded the 14th US National Securities Exchange (Long-Term Stock Exchange) with Eric Ries, where we built a fully regulated trading system from scratch. We started with 6 Raspberry Pis in a hotel room and scaled to production at a major data center in northern New Jersey where financial institutions colocate, maintaining 100% test coverage and rulebook-as-executable-tests throughout. Later, at Citadel, I led the team replacing their 30-year-old global position management system, serving the entire firm with modern infrastructure while maintaining the discipline that high-stakes financial systems demand. These experiences taught me that lean and rigor aren’t opposing forces—they’re complementary. You can build robust, enterprise-grade systems while keeping the process lean, answering critical questions upfront to prevent costly technical debt later.

Now I’m applying that same philosophy to AI development. I’m building production RAG systems, highly scalable low-latency agentic systems, and experimenting with self-hosted infrastructure to understand real economics and performance tradeoffs. The AI field is filled with marketing hype and tutorial-grade examples that ignore the observability, testing, and guardrails that financial systems solved decades ago. Through this blog, I want to cut through the noise and show how to build production-grade AI systems using proven engineering practices. Whether you’re working on regulated enterprise systems or personal projects, these practices ensure you get it right the first time.