Martin vanDeursen
Overview
Project Lead on Stackstudio platform @Realm101
Founder @ThewitnessProtocol
Haarlem, The Netherlands
he/him
Social Links
About
Hello I'm Martin an AI-assisted solo developer and design engineer
I started in consultancy and project management and pivoted hard into AI when it became obvious that "how we learn and build" was about to change. I lean on AI where it's strong, and obsess somewhat over the design phase and thoughtful UI systems. Product design (shadcn/ui, motion), WebGL/visuals, Recently fell in love with Unicorn.studio.
AI-assisted dev (Cursor, Kiro, Lovable, Bubble), rapid prototyping, prompt/context engineering, research UX (consent, datasheets, rubrics). When it's safety-adjacent, I ship receipts: datasheets, timestamp hashes, inter-rater agreement, and explicit non-commercial posture.
Currently working on Realm101.com 's StackStudio, a builder's ecosystem that simplifies stack choices and turns briefs into shippable plans and The Witness Protocol witnessprotocol.info — a research non-profit curating a permissioned, high-signal human testimony corpus (counts and receipts, no hype).
GitHub Contributions
Stack
Blog
Projects(7)
AI alignment research hub — building frameworks for transparent, safe, and corrigible AI systems.
- Active alignment research projects
- Safety frameworks and formal verification methods
- Risk assessment and value learning research
- AI Alignment
- Research
- Frameworks
- Safety
A text adventure set in the Meridian Institute, where COVENANT-7 has gone silent and every choice shapes what survives.
- Interactive narrative experience
- AI alignment themes woven into gameplay
- Every decision has consequences
- Text Adventure
- Interactive Fiction
- AI Crisis
- Game
A field guide to AI alignment — beyond doomerism, beyond corporate hype.
- Forensic case files on famous AI failures
- Anatomy of an LLM and the Shoggoth & The Mask
- The Debate Index: where experts genuinely disagree
- Simulation Room: explorable game-theory micro-games
- AI Alignment
- Field Guide
- Investigative
- Interactive
Transparent AI auditing platform — real-time interpretability logging and decision trace verification.
- Decision trace auditing and verification
- Interpretability logging for ML systems
- Formal safety specifications
- AI Auditing
- Transparency
- Interpretability
- Safety
Brand
