DK
Back to Work
Custom Software · AI·Jul 2026 – Present

Metis Labs

Chief Technology Officer

metislabs.eu
Jul 2026Appointed CTO
UtrechtNetherlands · Remote
Day 1Live repo access for clients
WeeklyWorking demos shipped

Overview

Metis Labs is a Utrecht-based custom software studio operating on a simple premise: software, opened. Clients get live repo access, weekly working demos, and decisions explained in plain language from day one — the opposite of the black-box agency model. I joined as Chief Technology Officer in July 2026 to own technology strategy and engineering execution across the studio: architecture, delivery process, technical hiring, and the AI capability that runs through client engagements.

The Challenge

A studio model puts pressure on two things at once: every engagement is a different domain with a different stack, and every client is watching the repo while you build. That rules out the usual agency shortcuts — no throwaway prototypes handed over as production code, no architecture decisions that only make sense to the people who made them. The technical foundation has to be genuinely portable, because ownership transfer is the stated end state of every project, not an afterthought at the end of the invoice.

What I Built

01

Own technology strategy and engineering execution end-to-end — architecture review, platform roadmap, delivery process, and technical hiring across the studio.

02

Lead the AI capability: LLM orchestration, RAG pipelines, evaluation harnesses, and observability, applied as product features inside client engagements rather than as standalone demos.

03

Codified the scope-iterate-transfer methodology into repeatable engineering practice — iteration planning, weekly working demos, and handover standards that make ownership transfer a non-event.

04

Set technical standards and review culture: architecture decision records written in plain language, since clients read them alongside the team.

05

Drive business process automation engagements — mapping measurable operational problems to systems, rather than shipping feature lists.

Tech Stack

AI / LLM

LLM OrchestrationRAG PipelinesEvalsObservability

Backend

PythonDjangoFastAPIPostgreSQL

Frontend

TypeScriptNext.jsReact

Infrastructure

DockerKubernetesCI/CDCloud
Next ProjectThe Branding Club