5 of 8 · ~2 min read
The people
A small project, but the roles matter. Solo on the technical side; distributed on the relational side, with very explicit boundaries so things don't fall through the gaps.
Kevin Hill — project lead
Independent researcher based in the US. Built the technical system end-to-end: model selection, ensembling, post-processing, validation, demo. Also handles strategy, English-language stakeholder communication, and overall direction.
Doesn't do: Spanish-language outreach, paperwork, in-person calls in Asunción. Those are out of scope for me to do directly — and the project is now bottlenecked on them.
Fran — meteorological collaborator
Ex-DMH meteorologist, based in Asunción. Provides domain credibility, signs off on technical claims that go to government, and gives the project meteorological legitimacy when it's presented to Paraguayan institutions. Tuesday 7pm sync is sacred; everything else is best-effort.
What he doesn't do reliably: follow through on operational commitments — calls he agrees to make, intros he promises. That's not a personal failing; it's a mismatch between his role (busy ex-government meteorologist with a day job) and the role this project ended up needing (full-time operator). The plan accounts for this honestly: he stays as the technical co-signer, and the operational work moves elsewhere.
Bilingual research analyst — to be hired
The role being filled in Track 1. Spanish-speaking, based in Asunción, with prior experience in Paraguayan government, NGO, or research environments. Hired through Workana on a 90-day contract with a paid trial week.
Their job: draft formal Spanish letters, make phone calls to government offices, schedule and attend meetings, build a contact list, file the DMH waiver request, follow up consistently, send a written status report every Friday in English.
Not their job: any technical work on the model, sales/business development, financial transactions, long-term employment commitment.
How decisions get made
| Decision type | Who decides |
|---|---|
| Technical claims, model architecture, validation methodology | Kevin (with Fran reviewing public-facing claims) |
| Outreach language, target lists, meeting scheduling | Analyst (with Kevin reviewing letters in English) |
| Spending over $200 | Kevin alone |
| Anything that commits Fran's name or credentials publicly | Fran approves explicitly each time |
| Strategy and prioritization | Kevin (informed by what stakeholders ask for) |
How communication works
- Friday 30-min sync between Kevin and analyst, in English, status review and decisions.
- Tuesday 7pm sync between Kevin and Fran, domain consultation.
- Slack / WhatsApp for ad-hoc questions.
- Written weekly status report from analyst every Friday before the sync.
- Google Drive folder for all shared documents, contact lists, drafts, meeting minutes.
Why the structure is what it is
A solo founder + one part-time consultant + one paid contractor is a fragile structure. It works for 90 days because the goal over those 90 days is itself fragile: open four institutional doors. If three of them open, the structure has earned the right to scale. If none of them open, the structure was wrong and we'll learn what the real shape needs to be.
Next: cost & time — the actual numbers, what this is taking and what it would take to extend.