Named clients and specific implementations sit under nondisclosure. The engagements themselves can now be described at sector level, and are — each family below has its own entry. What stays out is who, not what.
Twenty estates at once, not one for nineteen years
The City of Jacksonville was one estate, held for nineteen years, known to the level of which patch panel a fiber landed on. This portfolio was the inverse: 88 numbered engagements across two years, twenty-plus of them live at any moment, across four sectors that share almost no assumptions, in mid-market estates of hundreds to low thousands of staff.
That inversion is a different job, and it is worth being explicit about which skills it does and does not exercise. Nobody holds twenty estates in their head. What can be held is a design language — the patterns, the review practice and the standards that let fourteen architects produce work a client cannot tell apart by author. Depth in any one engagement comes from the architect embedded in it; consistency across all of them is the practice’s job, and that was mine.
Mid-market changes the constraint too. In an estate of nine thousand staff there is a specialist for everything and the difficulty is coordination. In an estate of eight hundred there is not, so a design has to be supportable by a team that will never be as large as the one that built it. A pattern nobody left behind can operate is a failed pattern, however elegant.
The engagement families
- Casino surveillance network design and segmentation, and a four-property gaming group
- SD-WAN across banking and a state legislature
- Healthcare estates across five providers, and a firewall refresh met by an intrusion the following night
- A diversified industrial manufacturer, across two acquisitions
- A natural gas and electric utility, from SCADA statement of work to delivery
- State and county government delivery, and a municipal assessment written to become the client’s own RFP
- Roughly fifteen E-Rate funded programs across school districts, an education service agency and a library consortium
- The seven-partner vendor program behind the capability the practice did not hold itself
- The post-quantum research program, the orchestration proof of concept and the practice lab they both ran in
- And one that did not close: the enterprise agreement pursuit for a global automotive manufacturer, lost, and followed by a headquarters meeting anyway
Four kinds of engagement
Design and architecture for a specific client build — data center, campus, WAN and security domains.
Assessment and current-state review, which is the work of saying what an estate actually is rather than what its diagrams claim, and what it can carry before anything is committed. The cryptographic inventory work and the AI readiness assessment elsewhere in this record are the same discipline on named subjects.
Delivery and implementation, because an architecture that is handed over at the design review and never seen again is a document rather than a system. Staying through the build is what closes the loop between what was drawn and what got installed.
Pre-sales and solution design — the architect in the room while the work was still being scoped, translating technical trade-offs into commercial terms for people deciding whether to buy. Being present before the sale is what stops an engagement from starting with a commitment nobody could design against.
Senior technical authority and executive escalation for enterprise accounts ran alongside all four: a player-and-coach arrangement in which design direction, escalation and hands-on architecture sat with the same person.
Why the count matters
Twenty-plus concurrent engagements against ten to twelve working architects is roughly two each at any moment, in four sectors, at a services provider where the client owns every decision and none of them report to you. Eighty-eight across two years is the same fact seen along the other axis: engagements that start, run and close continuously rather than a fixed set worked to completion.
It is the same constraint as the federated city estate, arriving from the opposite direction — and it is the reason the reference architectures, the standards and the working design review described in the architecture practice were built first and defended afterwards. Without them a portfolio that size delivers twenty different opinions and calls it a service.
Outcome
One design language across a portfolio no single person could hold in their head, which is the only thing that makes twenty concurrent engagements survivable.