Welcome
Hi, I’m Albert. I’m a developer at ePublic Solutions, building software for Dutch government and enterprise clients. Most of my work revolves around AI integrations, and the infrastructure that makes them reliable in production.
This blog is where I write about what I learn along the way: building tools around LLMs (MCP servers, CLI filters, hooks), making AI-assisted development actually work (guardrails, playbooks, structured workflows), and the engineering fundamentals that don’t change regardless of what technology is popular this week.
I have a love/hate relationship with AI. It can be brilliant and stupid within two turns of the same conversation. But when you give it the right tools and a clear playbook, the results can be genuinely impressive. Most of what I write is about finding that balance.
I’m also available for freelance consulting and development. If you’re working on AI implementation, architecture, or tooling and want to compare notes or work together, reach out below.
Blog
-
When LLMs Actually Deliver
-
gtk: Filtering CLI Noise to Save LLM Tokens
-
Security by Design: Using Project CodeGuard as AI Guardrails
-
Evidence-Based Best Practices as AI Guardrails (Part 2)
-
Evidence-Based Best Practices as AI Guardrails (Part 1)
-
Bijna de helft van alle bushaltes in Nederland is ontoegankelijk, dus bouwde ik een tool om er wat aan te doen
-
Let the AI Pick React
-
Vibe Coding: Product Quality and Democratisation
-
Securing YOLO Mode: How I Stop Claude Code from Nuking My System
-
Reverse-Engineering Figma Make: Extracting React Apps from Binary Files
-
Dictator: A Push-to-Talk Speech-to-Text App for macOS
-
Rethinking Claude Flow: From Per-Repo Chaos to Global App
· subscribe via RSS
Frequently Asked Questions
Practical AI development: building MCP servers, Claude Code customization (hooks, skills, guardrails), production hardening, Python, security, and the craft of making software reliable. I write about what I actually build and use, not theory.
AI is a tool, not a replacement for thinking. LLMs work best when you give them structured tools and clear playbooks, not when you let them improvise. Engineering fundamentals still matter. Sometimes the right answer is a database query, not an AI.
Not typical: intensive care nursing, political science, then software engineering. Healthcare taught me that systems must work under pressure. Political science gave me a lens for how organizations actually function. I’ve been programming since the early 90s, but took the long way round to making it my job.
Yes. I work as Lead Developer at ePublic Solutions, but I’m available for freelance consulting and development. Architecture reviews, proof-of-concept builds, AI implementation, and technical advisory. Remote across Europe, occasional on-site in the Netherlands.
Get in touch
Interested in working together? Let's start a conversation.