Founder of Ninobyte. Connected AI Operator building governed AI workflows, proof-of-work systems, and education infrastructure for emerging markets.
Connect. Govern. Execute. Prove.
I build AI systems that do real work safely. As founder of Ninobyte, I design and operate governed AI workflows — systems where AI reads first, writes only with approval, and produces portable proof of work. My focus sits at the intersection of cloud operations, AI tooling, product execution, and the practical operator workflows that connect them.
Now: shipping Prepnest, an offline-first WASSCE Study OS for Ghana (private, pre-release), and running the AI-Native CloudOps Lab and AI Security & Governance Lab on AWS.
| Project | What it is | Stack | Status | Proof |
|---|---|---|---|---|
| Prepnest | Offline-first WASSCE Study OS for Ghana — bundled study library, local search, governed content release gates | Expo · React Native · SQLite · Manim |  |
Docs · Case Study |
| Ojanaa | WhatsApp-native POS with invoice trust loop for emerging markets | TypeScript · React Native · Supabase |  |
Docs · Case Study |
| Ninolex | Pronunciation infrastructure for AI voice applications | TypeScript · Python · Next.js | |
Docs · Case Study |
| Ninobyte | Connected AI Operator doctrine — three-agent local AI operating system, governed workflow practice | Claude Skills · MCP · Python |  |
Docs · CloudOps Lab · Security Lab |
Product application code is private by design; each project publishes architecture docs and case studies instead. Prepnest is an independent product — not affiliated with or endorsed by WAEC.
| Project | Description | Stack |
|---|---|---|
| Circle MCP | AI operating interface for Circle.so — community operations through natural language. 16 tools, test-backed. | TypeScript |
| Ninolex-GH | Open Ghanaian pronunciation dictionary for TTS and AI systems. IPA, PLS, JSON. | Python |
| Sanity Education Starter | Archived — CMS schema starter from Prepnest's earlier architecture generation; preserved as a reference. | TypeScript · Sanity |
- Connect. Govern. Execute. Prove. Every workflow connects AI to real systems, governs writes with approval gates, executes against a clear outcome, and ships a portable proof artifact.
- Read-first AI. Default to read-only access; promote to write only after a human approves.
- Data model first, then features. The schema is the foundation.
- Runbooks, decision logs, and audit trails are first-class deliverables — not afterthoughts.
- Phase-gated development. No Phase 2 until Phase 1 passes all quality gates.
Languages — TypeScript · Python · SQL Frontend — React · Next.js · React Native · Expo Backend — Node.js · Supabase · PostgreSQL · SQLite AI / Voice — Claude · ElevenLabs · Vapi Infrastructure — AWS · GitHub Actions · Modal · EAS
If my open-source work is useful to you, consider sponsoring on GitHub. Contributions help fund maintenance, documentation, and new tooling.
Full portfolio → — case studies, architecture, ADRs, runbooks, proof packs