I build software that builds software.
Autonomous AI agents that research, verify, code, review, and publish ·
every step auditable, every decision traceable, humans always in control.
On GitHub since 2009 · 19,000+ contributions in the last year · 400+ pull requests across 35 repositories.
Autonomous research and publication systems. AI agents that do the research, verify every claim against original sources, and carry the result through the full publication workflow. The entire chain is auditable: every fact traces back to a source, every action to an approval. Humans stay in charge through intuitive approval workflows and delegated access control · you decide what runs on autopilot and what waits for sign-off.
An AI-first development factory. With the Prospergenics team I have streamlined scrum, DTAP, CI/CD, and devops around agent-driven development until the cycle time collapsed: a deliverable goes from request to production in hours to days, not weeks to months. Multi-agent systems write the code in isolated worktrees, open pull requests, review each other's work, and deploy · under hard safety rules and human approval gates.
Protocol and platform engineering. Ported the WhatsApp Web protocol to .NET (dawa, a full C# port of Baileys with no Node.js dependency), built LLM application frameworks, enterprise IAM, and production systems for clients in provenance research, water treatment, and e-commerce.
This development factory is available for your product. If you want deliverables in days instead of quarters, with a fully auditable AI pipeline behind them: info@martiendejong.nl
These are principles of decentralized cooperation: how autonomous parties work together fairly without a central authority. They are technology-agnostic, which is why they work here · people, organizations, and AI agents cooperate under the same seven questions:
| Principle | The question | |
|---|---|---|
| 1 | Reciprocity | Do I treat others as I expect to be treated? |
| 2 | Reasonableness | Would an impartial person consider this decision reasonable? |
| 3 | Proportionality | Is my response proportional to the situation? |
| 4 | Consent | Are all affected parties free to agree or refuse? |
| 5 | Accountability | Who is responsible for this decision, and is it traceable? |
| 6 | Non-Aggression | Does this action avoid causing unnecessary harm? |
| 7 | Non-Neglect | Am I ignoring someone or something that depends on this decision? |
The first four establish fairness and trust; the last three guarantee traceability and prevent harm through action or omission. Every agent workflow, approval gate, and business decision in these systems is evaluated against this sequence.
| Project | What it is |
|---|---|
| dawa | C# port of Baileys: WhatsApp Web client library for .NET, no Node.js required |
| whatsapp-bridge | WhatsApp HTTP bridge powered by Dawa · ASP.NET Core |
| Hazina | C# framework for LLM-powered applications and agents (formerly DevGPT) |
| autonomous-dev-system | Battle-tested autonomous agent protocols for Claude Code: worktree management, zero-tolerance enforcement, 75+ documented patterns |
| collatz_ai | AI-assisted mathematical research on the Collatz conjecture, with Lean 4 formalization |
| iam-system | Enterprise IAM and smart-building management: OAuth2/OIDC, RBAC, multi-tenancy, IoT integration |
None of this is a solo act. These systems are built and carried every day by a team of engineers in Kenya, working under the Prospergenics banner. The projects are AI and human work combined: the point is to help people, and to build beautiful things. My role is support: mentorship, architecture, and infrastructure. The shipping is theirs.
Diko Mohamed · Founder of AgentX, her own AI-agent company and implementation partner of Lua. She runs her own client projects alongside our shared work, from Python data projects to AI-driven SEO automation. The most autonomous engineer I work with.
Frank Kobaai · Agent-framework engineer. Frank builds the config-driven automation daemons that power several of our production AI research systems, and ships complete platforms of his own, like Rena Farm, an agricultural management platform in React and Node.js.
Lesiamon Mpoe · Developer and QA lead. Lessy is the quality gate of our pipeline: nothing reaches production until it survives his testing. Alongside QA he builds his own web platforms and AI experiments.
Sandra Mpoe · Developer. Sandra builds client websites end to end, like Oltepesi Safaris, and is expanding fast across the AI development stack.
Most of my work happens in private repositories: an autonomous agent ecosystem generates and reviews code around the clock. The public repos above show the tooling that makes that possible.
Open to collaboration on AI agent systems, provenance research, and .NET protocol work.
📫 info@martiendejong.nl