Operations engineer. Software builder. Liverpool, UK. Two disciplines in one head, fluent in both.
Trained as a manufacturing engineer. Since then I’ve led design work in industry, built predictive-maintenance ML models on real production sensor data, and run discrete-event simulations that surfaced material cost reduction in plant operations.
The toolset that bracketed that work: process, throughput, KPI, benefit sizing, simulation, predictive analytics, lean on the operations side; Python, FastAPI, Next.js, TypeScript, and ML models against live operational data on the engineering side. Both kept current.
I started Valoren Digital because there’s no good buying option for independent consultants who need their methodology in software. Generic dev shops can’t model your calculations. Consultancies can’t ship clean code. The gap is wide enough that consultants spend hours of QA correcting tools someone else built. Valoren sits in that gap so you don’t.
Brief a generic developer on a benefit-sizing calculator and you spend the first half of the engagement explaining what a benefit case is, why the methodology has the assumptions it does, and which edge cases make a number wrong in a defensible way. By the end of that explanation, you’ve done half the work and the developer has built half the wrong thing.
Brief me on the same calculator and I’ve already had those conversations. I’ve done the simulation work. I’ve watched a plant manager cross their arms when a number doesn’t match their gut. I write the code and I read the calculations the way an operations practitioner would.
That’s the difference. Not magic, not marketing. Two disciplines in one head: operations and engineering, fluent in both.
01Live engagement
Independent consultant · NDA
First commercial Valoren engagement in clients' hands. Same shape every other engagement takes.
02Award-winning research
Predictive ML
12+ hours of warning before CNC tool failure on production sensor streams. Full technical write-up below.
The architecture and data pipeline below are the shape Valoren delivers inside a Predictive-models engagement when a consultant’s methodology needs early-warning intelligence.
Research artefact, not a client deployment · 2025
A vibration + thermal model that predicts CNC tool failure 12+ hours before breakdown, hitting 99% accuracy on production sensor data. The pattern (operational signal in, early-warning artefact out) is what gets deployed inside a Predictive-models build for clients whose methodology hinges on getting ahead of an event.
Async-first, UK-based, 4 to 8 weeks per engagement. One active engagement at a time so the live client gets full focus. Weekly written updates, ad-hoc messaging on Slack or email, video calls only when needed and always with a written summary after.
No subcontracting, ever. One named senior builder per engagement, owning methodology absorption, spec, and build end-to-end. Subcontracted operational logic is where wrong calculations come from. The principle holds as Valoren grows.
Free 15-minute intro call to check the fit. No pitch, no work product. If there’s a fit, the next step is a paid 90-minute scoping session that produces a one-page tool spec and pricing recommendation.