Why this exists
The portfolio behind the portfolio
01
Get back hands-on
After years leading architecture at MultiChoice and Nedbank, I wanted to verify something: could I still build the systems I was designing? Not prototype-quality. Production-grade, with tests, observability, CI/CD, and the patterns that make things reliable at scale.
02
Rebuild my payments domain
I'd led payment platforms across mobile money, card acquiring, and merchant integrations. I rebuilt the core pieces from scratch using modern Java: a billing foundation first, then an M-Pesa event-driven gateway, then a reusable payment gateway. Each project feeds the next.
03
Test design-first AI development
I used the M-Pesa integration as a controlled experiment: write the full architecture document before touching code, then use AI tooling against that spec. The design document served two audiences: the human reviewer and the AI coder. It worked. 3 weeks of AI-assisted development delivered exactly what the spec required. Code review remained the critical skill throughout.
04
The honest finding
Without strong architectural judgment and code review skills, AI-generated code drifts. The specification, the review discipline, and the iterative correction. That's where the 20 years shows.
05
Get certified
The home lab proves the concepts. AWS Solutions Architect maps them to cloud-scale infrastructure in a language enterprise hiring managers recognise. SAA-C03 is next.
06
Harden the infrastructure
No load balancer. No database replica. No failover. That is honest: it is a portfolio home lab today. The next phase adds PostgreSQL primary/replica, Nginx upstream load balancing, and proper HA. The concepts page needs the infrastructure to back it up.