Build the layer that lets leadership lead.

Every hire routed through executives who were simultaneously fundraising and building partnerships. There was no management layer beneath them, so growth and leadership availability were in direct competition. Department heads had been "hiring" people by having them start. Priorities didn't exist, because nobody had time to set them.

The ask was never "hire people." It was: build the layer that lets the leadership leave the building.

A department-lead layer. That was the structural move — headcount demand started coming from the people doing the work instead of from executives guessing. Everything else hangs off it.

Then the machinery: a three-stage pipeline with role-specific skills tests, the same shape every time. An intake form that captured need, scope, and start date before anything moved, routed to the COO for budget. An automated signing packet — one email address in, every document out, the accountant and the executives notified on completion. A cohort onboarding model where everyone hired in the same two-week window met the CEO together, so she wasn't sitting ten separate introductions while the company hired weekly. A weekly all-hands with new faces every week. Payroll on Gusto and bill.com for a large international contingent. Monthly per-investor reporting.

12 to 43 in 12 weeks, because hiring decisions stopped waiting on executive calendars.

Intake ran bilingually — screens in English, answers accepted in either language — because the mandate was to source in Latin America and the U.S. latino community, and that rules out a plain job-board search. The system was a filter with a budget attached, not a faucet: every request was checked against level, contractor-versus-employee, and the actual envelope.

The founders opened and closed every senior offer. The fundraise wasn't mine. Some hires bypassed the system entirely and were cleaned up after the fact.

As of September 2026: they kept my steps and dropped my sequences. The process survived; the automation depended on me and didn't. The department-lead layer never fully took — twelve weeks wasn't long enough for a company that had never had management to own it. The lesson I carried forward is that transferability is a design requirement, not a documentation task. A runbook doesn't make a system portable if the system needs judgment to operate.