I build AI-native products with small teams that think big.

Freelance full-stack engineer and product lead in Amsterdam. I take AI ideas from prototype to production — and love working shoulder-to-shoulder with founders and small teams who move fast.

◷ Amsterdam · CET ◐ Python · TS · React · Svelte ◆ Product · AI · Data
arnoud@ams — ~/portfolio
arnoud@ams ~ %
01 / what i do
01

AI-native product building

Products designed around AI from day one — not features bolted on later. From the core idea to the shipped experience your users actually touch.

02

Full-stack product engineering

End-to-end features — React or Svelte front ends on Python and Node back ends, with Postgres or ClickHouse underneath.

03

Product-led prototyping

Fast, credible prototypes that turn a vague AI idea into something you can put in front of users and learn from this week.

04

Agentic & autonomous systems

When a product needs to plan, act and self-correct on its own, I design the agent loops — and the guardrails that keep them trustworthy.

02 / selected work
03 / the lab

Ongoing experiments — prototypes and tools I build to stay sharp and de-risk client work before it ships.

harness/ active

A toolkit for building eval + guardrail harnesses around LLM agents — deterministic replay, scoring, and regression tracking.

agent-playground wip

A sandbox for autonomous tool-use loops: watch an agent plan, act, fail and self-correct in real time.

clickstream active

Realtime product analytics on ClickHouse — sub-second queries over billions of events.

svelte-sketches ongoing

A rolling set of tiny interactive prototypes exploring novel product UI ideas.

04 / the stack
Languages
Python TypeScript JavaScript SQL
Frontend
React Svelte Vite Tailwind
Backend
Django FastAPI Node Postgres Redis
Data
ClickHouse Elasticsearch Kafka Postgres MongoDB dbt
AI / Agents
LLM agents Tool-use loops Eval harnesses RAG
Infra
Docker Kubernetes AWS GitHub Actions ArgoCD Fly.io
portrait.jpg
05 / about

I'm Arnoud — a freelance product engineer who likes turning AI ideas into products people actually use.

Most of my work is AI-native product building: taking a rough idea, shaping it into something real, and shipping it with a small team. I care as much about the product decisions as the code. Across ~15 years I led engineering at Cineville in Amsterdam, consulted on data-heavy backends for Alliander and Sendcloud, and now lead development at Penelope, a New York fintech.

I'm equally comfortable in a React/Svelte front end, a Python or Node back end, and a query planner over ClickHouse or Elastic. I do my best work in small teams that think big — honest scope, tight feedback loops, and code that keeps running when I'm not looking.

15 yrs
shipping product
70k+
subscriptions built
20×
faster grid registry
06 / let's talk

Let's build a product worth shipping.

AI-native product builds, full-stack engineering, and prototype sprints with small teams that think big. Tell me what you're making.