A multi-year, ongoing build with fairkom: a news aggregation and distribution platform for independent European journalism and broadcasters. Lead developer and technical project management — full-stack delivery, requirements work, code review for the team, and an easy to deploy and use platform.
Now into its third year — a public-funded platform built to outlive its own funding cycles.
A distributed job pipeline built inside Display Europe to coordinate long-running work — machine transcription, translation — across a heterogeneous toolchain. Jobs arrive via a Node-RED REST interface (proxied through Kong), get persisted and distributed over Kafka, then picked up by distributed connectors that follow a small plugin contract. Responses happen either inside the request-response cycle or via Webhooks.
Adding a new transcription or translation backend is a connector, not a rewrite.
REPCO is Display Europe's main upstream — variable in both response and content. The caching framework wraps it for predictability: Celery and Redis parallelise requests and form a short-term cache with built-in invalidation; pydantic validates schemas as data arrives; PostgreSQL holds the long-term cache. A write-behind pattern accepts brief stale-read windows in exchange for stable response times. Communication via REST and a Redis pipeline.
The frontend gets a stable contract; the trade-off — short windows of stale data — was a deliberate one, named in the design.
Hosting for web applications and email, currently being rebuilt to reflect the political and social moment we're in: sovereignty as a working assumption, encryption wherever possible (email at rest included), and tools that small teams and offices can actually use day-to-day. The rebuild is being documented as it happens — livestreams, video notes, articles — as accountability and as teaching.
Built in the open, documented as it's built. The first principle is that you can outlive the provider — including, eventually, me.
"Below the threshold of toxicity.
No bloat. No theatre.
I solve the problem and explain the trade-offs I couldn't avoid."
I share rates, timelines, and uncertainty up front. If I don't know, I say so — and I tell you what I'd need to find out.
Distributed systems are a tax, not a feature. I reach for them when the workload demands it — not before.
Every engagement ends with a runbook, an architecture note, and an invitation to break the dependency. Lock-in is not a business model I want.
Open-source first, digital sovereignty as a working assumption. The systems I build can be inspected, forked, and outlived — by you, by me, by whoever comes next.
A media-arts background means I sweat what other engineers don't — naming, structure, the shape of the system as it ages. Speed is downstream of clarity.
Tell me what's broken, what's stalled, or what you'd build if you had a senior engineer on the bench for a quarter.