Writing
Writing
Essays on the systems I build and operate — production AI, agentic workflow systems, real-time data platforms, growth engineering, and the operating decisions that compound. Opinionated, written from the field, no filler.
-
Build vs. Rent: When Proprietary AI Infrastructure Becomes A Strategic Asset
Most companies should rent AI capability until they hit a specific volume and data threshold. The math, the inflection points, and the failure modes of premature in-housing.
-
Agentic Workflow Systems In Production: A Field Report
Demo agents and production agents are different species. The five things that break, the architecture that survives, and where agents actually win.
-
Real-Time Data Platforms Are A Moat. Most Companies Treat Them Like A Cost Center.
The latency hierarchy, why most "real-time" stacks aren't, the five things a real platform must do, and where the value actually lives.
-
Full-Funnel Growth Engineering: How To Stop Lying To Yourself About What's Working
Why dashboards lie, what a real measurement architecture looks like, the four numbers that actually matter, and the one report I make every operator build.
-
An Operator's Framework For Evaluating LLM and Emerging-AI Tools
Most operator AI evaluations are vibes. A working framework that forces you to test against your data, your latency budget, and your cost ceiling.