Architecture is not just about boxes and arrows on a whiteboard. It’s a design discipline — one that requires the same rigor and intentionality we apply to any creative work.
The Design Mindset
When I built KalmSkills — integrating O*NET, SEC EDGAR, and BLS data into a single career intelligence platform — the architectural decisions weren’t about which framework to use. They were about which federal APIs to trust, how to handle data staleness, and when to show users “we don’t know” instead of a stale number.
The best architectures are obvious in hindsight. That’s the mark of good design — it feels inevitable.
Making Trade-offs Explicit
Every architectural decision is a trade-off. The Program Viability Engine could have been a comprehensive platform. Instead, it’s a focused calculator that does one thing well: model whether a workforce education program will break even. That constraint was the architecture. The same constraint-as-design principle applies to the Workforce ROI Dashboard, where an API-first architecture ensures the dashboard is just one consumer of a general-purpose computation engine.
The discipline lies not in avoiding trade-offs, but in making them explicit, documented, and reversible where possible. This is where technical writing becomes architectural practice — a principle I explore further in The Case for Boring Technology. The same argument — that the bricks are cheap and the design of the building is not — appears in a very different register in Doing Academic Philosophy in the Age of AI, where the philosopher becomes a systems architect rather than a text producer.
Decision Records
Architecture Decision Records (ADRs) are perhaps the most underrated tool in a system designer’s toolkit. They capture not just what was decided, but why — including the alternatives considered and the constraints that led to the choice. The same discipline of making thinking visible is what distinguishes efficient AI collaboration from expensive meandering — a lesson I document in On Finite Tokens and Infinite Tasks.
Related
Projects: KalmSkills · Program Viability Engine · Workforce ROI Dashboard · Observability Platform
Writing: The Case for Boring Technology · The Case for Boring Automation · Doing Academic Philosophy in the Age of AI · On Finite Tokens and Infinite Tasks
