I build master data platforms that give everyone the same answer,
and deployment pipelines that no longer depend on any one person.
TIBCO EBX developer, .NET and microservices architect, DevOps. Sixteen years on systems that can't afford to get it wrong.
One path to write.
Another to read.
On a large insurer's party master data platform, the two needs were in conflict: guarantee consistency on write, without keeping the reading systems waiting. Two contradictory requirements in the same system. The answer was to separate the two paths.
Two records. One company. Are you sure?
Identity resolution means deciding whether two records refer to the same entity. The problem is never the tool, it's the rule.
I designed and built the entire Dun & Bradstreet component of a party master data platform: API calls, data consolidation, matching logic on DUNS and Name-Address. The real work isn't calling the API — it's making the call on the ambiguous cases, and making every decision traceable and reproducible.
More than a dozen systems, producers and consumers, rely on that platform. Each one needs the answer to be the same.
What I do.
What I won't do.
Commitments
- Make the environment reproducible before optimizing anything
- Check that what worked before still works
- When a mistake comes back, fix what let it through
- Write down the decisions, not just the code
- Flag a blocker the day it appears, not at the next status meeting
- Hand over a pipeline the team can run without me
No-gos
- Generated code nobody has reviewed
- Stacking one more layer on a system nobody understands
- Applying a solution because it worked elsewhere, or because it's considered best practice, without checking that it holds here
- Claiming figures nobody has measured
- Master data whose rules live in one person's head
Developer first.
Architect second.
The signal was the projects I was passing up.
Fourteen years at Matmut. I really liked it there: a solid team, interesting work. The signal wasn't there — it was in the technical work I was passing up, the kind that would have thrilled me a few years earlier. Without noticing. It's a stupid indicator, but it works.
It wasn't running away. I needed a new challenge, and I chose to own that rather than let my motivation erode.
Then the first contract came through: full remote, in English, in an international team. The technology, the language, the working culture: everything was new at once. I had prepared for it — I'd put the time into EBX and into my English before starting. Looking back, it was exactly what I needed.
Hexarace
An ecosystem of SimHub plugins that I design, build and distribute on my own. It's my proving ground: on a product of my own, I try what has no place on a client project.
HexaDash
High-density dashboards, from the 4-inch DDU to the full-screen overlay. Over 80 sims supported through SimHub.
HexaShift
Steering-wheel LED control: engine revs, flags, alerts. Ten themes.
HexaLight
Ambient lighting synced to the race — Philips Hue, Govee, LED strips.
What I validate here
- Real time
- Process a continuous telemetry stream and keep the display smooth.
- Reading at speed
- Information that has to be looked for is information lost — and lap time lost with it.
- Distribution
- Version, publish and get a binary installed by people I don't know — then ship every update after that in one click, from the module itself.
- AI-assisted development
- This is where I make AI-assisted development production-grade. A multi-agent architecture where each agent has a defined role — specification, architecture, development, testing, review, documentation — with what makes it genuinely usable: systematic review, automated tests, decisions on record. I prove it out on my own product, and I know how to apply it where mistakes are expensive.
Master data to build,
or to set straight?
contact@hexainnovate.com
Remote or on site. Based in Normandy, assignments in French and in English.
A master data platform
To build, to take over after someone leaves, or to move to a new version without breaking the systems that depend on it.
A delivery pipeline
To automate delivery when it still depends on manual steps, or to take over when nobody knows what is running in production anymore.