Founder and engineer building across fintech, agtech, health tech, and creative technology for East African and Global South markets. I work founder-operator style — a lean, AI-native build process where I design architecture and make the calls, and AI agents handle a large share of implementation, code review, and infrastructure work under my direction.
| Project | What it is |
|---|---|
| GoPay | East African fintech super app — wallet, payments, government services, merchant tools. Built compliance-first: real ledger and idempotency infrastructure now, production payment credentials deliberately held back until Bank of Tanzania licensing is in place |
| Kilimo AI | AgTech platform for East African farmers — real-time market listings, farmer profiles, an income ledger, and a Swahili-aware assistant, running on Supabase with RLS-enforced data access |
| Kaya | AI-powered rental property management for Ontario landlords and tenants, built for the CHI Accelerator |
| Kultr Hub | Events and ticketing platform for East African markets, with mobile-money payment collection and a creator payout system |
| SEEN | Interactive multilingual storytelling platform — audio, film, and cultural content |
| Akili Core | Multi-agent automation platform running my own operations — scheduling, monitoring, and research agents built on Python/uv |
These sit under CREOVA, the umbrella I build them through — a multi-vertical tech ecosystem, not a single product. See the full repo list for the rest of the portfolio.
- Security and correctness before feature breadth. Financial and health-data products get RLS, real ledgers, and idempotency handling before they get polish.
- Honest product status over the appearance of progress. Demo-mode features are labeled as demo; a passing build is not the same claim as a tested one.
- Compliance-aware sequencing. Regulated products (payments, health) are engineered for their eventual regulatory bar without prematurely claiming licenses or certifications they don't yet have.
- AI-assisted, human-directed. Every repository runs on the same governance baseline — branch protection, PR review, and a
CLAUDE.mdwritten for how AI coding agents should actually operate in that specific codebase.


