Component reality
Slots, states, wrappers, accessibility APIs, and migration cost matter. Good systems do not pretend otherwise.
UX Engineer at Microsoft / Fabric ecosystem
I build interface systems that hold up under product pressure. Design systems, Web Components, and AI-assisted tools that reduce drift, remove guesswork, and make good implementation easier.
Working principles
Slots, states, wrappers, accessibility APIs, and migration cost matter. Good systems do not pretend otherwise.
Keyboard paths, visible focus, contrast, state behavior, and examples are product structure, not cleanup work.
Useful automation reduces ambiguity, names tradeoffs, and leaves a better trail than a person could make by hand.
Specs, examples, docs, and review notes should make the right implementation feel obvious.
How I work
I am a 35-year-old UX Engineer at Microsoft working in the Fabric ecosystem. I focus on component architecture, UI infrastructure, design systems, and AI-assisted development workflows. Most of my work lives between design intent and production code: tokens, components, wrappers, examples, validation, documentation, and the tools that keep all of it aligned.
Recently, I have been building systems that help AI agents understand product UI the way engineers need them to: with component constraints, accessibility expectations, framework patterns, repository context, and enough guardrails to avoid inventing a second design system in the shadows.
Away from the editor, I like kite surfing, playing pinball, playing music, and going to shows. I care about systems that are rigorous enough to survive production but still leave room for taste, attention, and a little human electricity.
Work
Manage a 160+ component Microsoft Fabric UX platform built on Web Components as the source of truth, with thin Angular and React wrappers for enterprise Power BI client migration work across Fabric Web and Fabric Copilot surfaces.
An agentic development layer for Microsoft Fabric UX adoption: secure Figma intake, evidence-based component parity, CLI-driven diagnostics, optional ADO issue filing, and gated code generation for React and Angular teams.
A deterministic component-health system for Microsoft Fabric UX that turned review checklist conventions into executable rules, resolving 5,000+ issues across exports, wrappers, Storybook, docs, metadata, and API drift.
A real-time workflow observatory built for Output.ai — virtualized execution feeds, step-level trace debugging, timeline swimlanes, cost analytics, and evaluator inspection. A prototype demonstrating how AI pipeline observability becomes human-navigable through design engineering.
An operator-facing trading terminal for portfolio accounting, strategy construction, AI-assisted planning, deterministic risk controls, broker-safe execution, backtesting, and production operations.
An embeddable AI-native terminal for web apps where shell output can become structured text, charts, tables, file previews, graphs, and LLM-ready insight without breaking the terminal flow.
Built and repaired Microsoft Fluent UI Web Components inside Microsoft's Fluent UI codebase: custom elements, Storybook examples, high-contrast behavior, keyboard paths, focus states, manifests, and the small edge cases that decide whether a component can survive real product use.
A contractor and vendor compliance app for tracking certificates of insurance, renewal status, document review, audit history, and the practical question property teams care about: can this vendor work or be paid right now?
A card-game probability workbench I built for a friend who wanted clearer deck-building decisions: paste a deck, tag cards, define the hand or board state that matters, then compare ratios with exact math and seeded simulation.
An older automation project with a simple rule: take two public text streams, define a recombination engine, and publish the result on a schedule. Built with Node, Cheerio, request, Twit, timed intervals, and a small text-generation loop.
A full-stack astrology workbench for exploring natal charts and relationship comparisons. Interactive chart wheels, aspect grids, and compatibility breakdowns make a dense domain inspectable, with a separate calculation engine and optional AI interpretation.
A browser-only Three.js AI art prototype and semantic weather map exploring memory, text, and generated atmosphere.
A local-first accessibility notebook for pedestrian observations, path checks, and non-driving trips, with explicit limits on what the data can promise.
Seattle discovery and day planning with source-linked places, itinerary editing, a day map, and three design lenses in light and dark modes.
A self-contained demo app for agentic workflow observability: run traces, static repository discovery, and an architecture graph, all tagging evidence provenance and confidence.
A fictional booking demo with guest, returning-patient, and staff flows, deterministic failure recovery, and FAST Web Components wrapped for React.
A token-driven React design system and the reference site that documents it from its own source: one token source resolving into three products, two themes, and two densities.
Writing
Contact
Best fit: teams building design systems, component libraries, accessibility-minded frontend infrastructure, or AI-assisted engineering workflows.