New book by Howard Riker
Reliable by Design
Engineering AI Systems That Hold.
A production engineering curriculum for software engineers who are done with demos and need to ship AI that survives real users, real budgets, and real 2 AM incidents.
// central argument
A model is a probabilistic kernel. Your software system must be the deterministic shell around it. That shell is where reliability lives — and building it well is an engineering problem, not a research problem.
Howard Riker
Head of Engineering at one of Europe's most-travelled consumer technology platforms. Howard has spent his career building teams and systems that keep working when everything else is on fire — and leading the transition into AI-native product development.
He has sat in the on-call rotation long enough to have strong opinions about observability, failure recovery, and the gap between demo-quality AI and production-grade AI.
Read full bio