Stephane Chevreux
IT Technical Leader & Design System Expert
14 years of experience in Front-end web app development. I architect React component libraries, lead Design System projects, and design AI-assisted development workflows — from context engineering and structured UI specs to Claude Code + MCP integrations that let agents generate production-quality UI.
Professional Experience
Entrepreneur in Design System & AI-Assisted UI
Freelance
Independent Front-end leader specializing in Design Systems and AI-assisted UI development. I build React component libraries, author structured UI specifications that serve as context for AI agents, and define the visual quality bar that AI-generated code must meet. I use Claude Code connected via MCP to Figma (figma-console) to inspect designs, extract tokens, and generate components from the source of truth — no copy-paste. Writing precise specs and system prompts (context engineering) is the core of the workflow: the richer the context, the better the AI output.
IT Technical Leader Design System
OVH
Architecture and creation of the OVH Design System in a monorepo (multi-package, multi-framework). Designed the semantic token architecture — colors, spacing, typography, radii — enabling cross-product theming and dark mode support by switching a single token layer. Built multi-framework compatible web components (StencilJs) consumed by React, Angular, and Vue apps. Supervised the developer team, defined Craftsmanship standards, and piloted RUN management with Scrum.
Lead Developer
Adictiz
Revamp and migration to Angular 8 of the SaaS campaign editing platform within a monorepo architecture. Built a shared component library using Angular Material and Angular CDK — the same model Google uses to craft Material Design: low-level CDK primitives for behavior, Material layer for visual styling. DevOps, code quality, and Design System study.
Lead Developer
Worldline
Design & development of the OPS internal platform for deploying and monitoring Worldline applications across all teams. Set up a monorepo with shared Angular components and a common design language, ensuring UI consistency across multiple product teams.
Software Engineer
Luxembourg State
Interface for managing road traffic offenses in Luxembourg. Smart push for EDF (Multi-channel push platform).
Software Engineer
French Government
Development of the PFLAU application for the French Government (Emergency Call Location Platform).
Software Engineer
Orange
Management of Orange messaging (voice and email). Deployments, SLA & quality of service, features, corrections, customer relationship. Development of the Orange Cloud on TV Boxes.
Full Stack Developer
SNCF
Design & development of an EDM application (Electronic Document Management).
Key Skills
Design Systems
5 yearsReact
4 yearsJavascript
14 yearsTypescript
6 yearsAngular
9 yearsLeadership
8 yearsUX Design
5 yearsTechnical Knowledge
React
4 years
Angular Material
5 years
Angular CDK
5 years
JavaScript
14 years
TypeScript
6 years
Angular
9 years
Node.js
10 years
Stencil
3 years
Jest
3 years
HTML5
14 years
CSS3
14 years
AI-Assisted Development Workflow
In an AI-first model, quality depends on the quality of the spec. I act as the bridge between design intent and AI execution.
Structured Specs First
Writing detailed UI specifications is the foundation: component API, visual states, semantic tokens, responsive breakpoints, accessibility. The richer the context fed to the agent, the better the output. Specs are not optional — they are the product.
Context Engineering
Crafting system prompts that encode the design system rules: token naming conventions, dark mode patterns, composition principles, do/don't. AI agents follow the spec, not the other way around. The system prompt is a living document, iterated like code.
Claude Code + MCP
Using Claude Code connected to Figma via MCP (figma-console) to inspect designs, extract tokens, and generate components directly from the design source of truth — no copy-paste. The agent reads the design, writes the code, and I validate the output.
Visual Quality Gate
Acting as the human quality guardian: reviewing AI-generated code against the spec, checking token usage, dark mode behavior, responsive layouts, and accessibility before any merge. This is where design expertise and engineering discipline converge.
Languages
French
Mother language
English
Technical
Education
Master's Degree
Jules Verne University, North of France
2005 – 2010
Specialized course in computer science and industrial systems. Internship at the MIS research laboratory. Sun speed webmaster for solar vehicles.
French HS Diploma in Science
LOUIS THUILLIER High School
2004 – 2005
Interests
Total renovation of my home with high insulation & KNX home automation (EIB, 1-wire)
SketchUp 3D modeling
Strong sensitivity in ecology
Shifter (The Shift Project)
Hiking and Jogging
Let's Work Together
Ready to bring your design system vision to life? Let's discuss how we can create exceptional digital experiences together.