Rijswijk, South Holland, Netherlands
Most AI coding is vibe coding. I'm building the solution against code slop. I'm the founder of Fallow, the codebase truth layer your coding agent can call. AI writes plausible code, but it has no reliable way to check whether a change made the codebase better or worse. So we paper over it with prompts like "be careful" and "act like a senior engineer". That is how you get AI slop: duplication, dead code, broken boundaries, and changes nobody fully trusts. Fallow puts engineering standards inside the agent loop instead of in wishful prompts. An agent, a developer, or CI calls Fallow through the CLI, MCP, or hooks and gets deterministic, reproducible answers about a TypeScript or JavaScript codebase: → what is dead → what is duplicated → what is risky → what actually runs in production The agent acts on those facts to refactor and self-correct before a human ever reviews the diff. The biggest productivity gains do not come from a smarter model or a better prompt. They come from a better harness: deterministic checks the agent can act on, and that a human can trust. Fallow's open-source core has passed 2,400 GitHub stars and over 700k monthly npm downloads, and it's already part of the daily workflow for developers around the world. The paid layer adds production-grounded evidence and team guardrails for organizations shipping faster with AI. AI made code cheap to produce and expensive to trust. Every team adopting coding agents inherits the same problem, and no incumbent owns the answer yet. That is the category Fallow is building. Featured on Syntax, the web development podcast by Wes Bos and Scott Tolinski, where Scott called it "the number one tool people aren't using" to fix AI slop. Alongside Fallow, I take on select freelance work as a full-stack developer and tech lead, helping teams adopt AI-assisted development without letting their codebase rot. Before Fallow I spent years scaling teams, shipping design systems used by large audiences, and building the architecture and testing practices that keep codebases maintainable. [email protected] · https://fallow.tools
I founded Fallow to fix a problem every team now has: AI writes code faster than anyone can review it, and most of that code ships with no reliable check on whether it made the codebase better or worse. Fallow is deterministic codebase intelligence for TypeScript and JavaScript. Instead of trusting a model to guess its way through a repository, developers, CI, and AI agents call Fallow through the CLI, MCP, or hooks and get deterministic, reproducible answers about the quality of the code. Engineering standards live inside the agent loop as checks the agent acts on and a human can review, not as wishful prompts like "be careful." What I do: • Set product direction, positioning, and the open-source plus commercial strategy • Build the full stack: the Rust analysis engine, the cloud backend (Bun + Elysia on Fly.io), the dashboard, the VS Code extension, and the MCP server • Grow an open-source community and turn it into a commercial product, Fallow Runtime The open-source core is free and already part of the daily workflow for developers around the world. Fallow Runtime, the paid layer, adds production execution evidence and team guardrails. Fallow was recently featured on the Syntax.fm podcast, where it was called "the number one tool people aren't using" to fix AI slop.
Freelance lead developer with a focus on front-end and solutions architecture. Experienced across modern JavaScript frameworks (React, Vue, Angular), native mobile development (React Native & SwiftUI), and full-stack delivery with Node.js, Laravel/Symfony, and headless CMS platforms. Advocate for Web Components, design systems, and accessible, performant web applications.
- Developing secure, accessible interfaces for reisdocument/paspoort application workflows across multiple portals using TypeScript, React/React Router, Next.js, and the NL Design System. - Implementing authentication and authorization across all user flows with Keycloak (OIDC), including guarded routes, session management, token refresh/expiry, and role-based access. - Integrating REST APIs with TanStack Query for server-state management (caching, background revalidation, pagination, retries, and error handling) across SPA and SSR/ISR scenarios. - Creating GitLab CI/CD pipelines for linting, type checks, build/test, and release, leveraging Vitest for unit/integration tests and Robot Framework for end-to-end coverage. - Delivering to WCAG AA through audited accessibility, embedding a11y practices in components, patterns, and documentation. - Collaborating with designers and stakeholders to adopt and extend NL Design System components for consistent UX across portals.
- Responsible for the user experience across all register frontends, translating complex form flows into intuitive, accessible interfaces in close collaboration with architects and product owners. - Developing a Vue 3 + TypeScript adapter library for JSON Forms that provides custom renderers wrapping the IND component library, used across ~10 register frontends for the Immigratie- en Naturalisatiedienst. - Building a form runtime composable managing the full form lifecycle: schema loading, Ajv validation (JSON Schema 2020-12), submission, and error mapping with Dutch localization via ajv-i18n. - Designing a ranked tester + renderer architecture that automatically selects the correct UI component for any schema/uischema combination, including complex nested contexts like array items. - Creating reusable composables for two-way data binding, validation state, required/readonly derivation, and wizard/categorization navigation across all form controls. - Developing Storybook stories for visual testing and documentation of all renderers, controls, and layout components. - Integrating client-side Ajv errors with backend 422 validation errors through a unified error pipeline keyed by JSON pointer path.
- Building the RotterdamPas mobile app (iOS & Android) using React Native, React Navigation, Redux Toolkit, TanStack Query, and Firebase for analytics and push notifications. - Developing the VPRO ClubLees reading app for children using React Native with the Immer reader module, Redux Toolkit, and TanStack Query. - Independently building the Port of Rotterdam Innovation Bridge — an interactive visualization tool featuring a Mapbox ecosystem map and strategic framework canvas, built with Next.js 16, React 19, styled-components, Jotai, and a Statamic/Laravel headless CMS backend. - Creating the Zeeuws Museum website as a monorepo with a Next.js frontend connected to Statamic CMS, featuring Algolia-powered search. - Building the Rotterdam Inclusief website from scratch using Next.js 15, React 19, the Rotterdam Design System (NL Design System), and Framer Motion. - Setting up CI/CD pipelines with GitHub Actions for linting, type checking, unit testing (Vitest), and E2E testing (Playwright) across projects. - Configuring and maintaining Statamic CMS backends built on Laravel, with Eloquent Driver, database migrations, PHPStan static analysis, and Deployer-based deployments.
Pioneering AI solutions for the world of tomorrow. We combine deep IT expertise with cutting-edge artificial intelligence to help businesses and governments work smarter, faster, and more successfully. - Retrieval-Augmented Generation (RAG) systems that combine private data with large language models for accurate, secure, and context-aware answers. - Chat-based AI interfaces that bring intelligence directly into everyday workflows — from customer service and internal knowledge bots to developer assistants. - Custom AI SDKs and agent frameworks, empowering organizations to integrate LLMs, embeddings, and reasoning capabilities into their own products. - Model Context Protocol (MCP) servers for seamless interaction between applications, APIs, and autonomous AI agents. - Advanced vector databases and data pipelines to ensure knowledge retrieval is fast, scalable, and always up to date. - Secure, compliant AI infrastructure, designed for enterprise and government environments emphasizing data privacy, auditability, and control. - Cloud-native orchestration and deployment pipelines optimized for AI workloads, including fine-tuning, evaluation, and monitoring. - Building IsAgentReady.com — a website scanner that scores AI agent readiness across 47 checkpoints in 5 categories (Content Discovery, Search Signals, Semantics, Agent Protocols, Security & Trust), built with Phoenix LiveView, Elixir, PostgreSQL, and Oban, deployed on Fly.io. - Publishing the @isagentready/cli npm package — a terminal tool that scans websites for AI agent readiness with human-friendly output and auto-detected JSON mode for piped usage. - Developing open-source Model Context Protocol (MCP) servers published on npm: anwb-mcp (traffic & routing), bol-mcp (Bol.com Retailer API), kvk-mcp (Kamer van Koophandel), postnl-mcp (shipments & tracking), spaceship-mcp (domain management), and isagentready-mcp (website scanning) — all built with TypeScript, full test coverage, and CI/CD via GitHub Actions.