EverythingI’ve shipped.

Discovery

I started with the category rather than the product. Every competitor photographs folded fabric on white; nobody photographs the person wearing it, which sells nothing to somebody facing a week in hospital. That audit set three things: the positioning, the two buyer segments — patients and the people caring for them — and the decision to lead acquisition with a free fall-risk assessment instead of a discount code.

Architecture

Static-first, deliberately. Hand-written HTML, CSS, and JavaScript on Vercel, with no bundler and no framework for the storefront proper. A catalogue this size does not need a build step, and skipping one buys near-instant loads and a site that will still build in five years without a dependency archaeology dig. The one genuinely interactive tool runs on hand-written React shims so it could not drag a toolchain onto everything else.

Design

Brand identity, palette, and typography, then the photographic direction — patients and caregivers in real light rather than product flat-lays — and a product-page template that had to carry seven very different SKUs, from a cotton gown to a wearable sensor.

Development

Every line is mine: the storefront, seven product detail pages, and the React fall-risk assessment that scores against WHO fall data with charts and a regional model.

Integrations & APIs

The assessment posts to a serverless lead-capture endpoint. Alongside it, a full technical SEO program — www unification, clean product slugs, an image pipeline, security headers, sitemap and robots.

Delivered

• Brand positioning, identity, and photographic direction from scratch
• Seven product detail pages — gowns, scrubs, compression socks, the AirGuard™ collar, MEND Pulse, MEND Oxi, and the MEND app
• React fall-risk assessment built on WHO fall data, wired to a serverless lead-capture endpoint
• Technical SEO program and a static-first architecture with no framework lock-in

Discovery

Every engagement opens the same way: current-state assessment. What systems exist, which of them believe they are the source of truth, and where the seams are. Most organizations do not have a technology problem so much as four systems that each believe they own the customer. Finding that seam is the billable part; closing it is the easy part.

Architecture

Target-state design, data modeling, integration strategy, and feasibility review — documented as technical context diagrams and as-is/to-be architecture so business and technical stakeholders align before the build rather than during it.

Design

This site and its apex parent are one design system deployed twice: a drafting-sheet identity, a single type scale, and an accent colour that tracks the visitor’s local clock. Hand-written, no framework, same reasoning I bring to client work.

Development

I stay on the tools through delivery, which is the part that keeps the architecture honest — an architect who does not build ships diagrams that cannot be built. Handover includes solution architecture documentation, integration specifications, and reusable design patterns, so the client’s team can extend the work without me.

Integrations & APIs

Multi-system event-driven integration automating project intake and ticket routing across Workfront, Slack, and Salesforce over REST; ERP and commerce data consolidated into Snowflake as a governed executive reporting layer, replacing spreadsheet reconciliation.

Outcomes

• Roughly 40% less triage handoff time from the Workfront–Slack–Salesforce integration
• Spreadsheet reconciliation replaced by a governed Snowflake reporting layer
• Architecture documentation and reusable patterns handed over for independent maintenance

Discovery

Three things had to work together, and each one undermined the others if built carelessly: an AI-powered Rate-to-Reveal game, the proprietary EyeQ+ scoring metric, and a global registry collectors would trust as a system of record. A grading platform lives or dies on whether people believe the number, so the discovery work was mostly about where that trust comes from and what would spend it.

Architecture

Cloud-hosted SvelteKit on Vercel. I architected the API integrations and workflow pipelines connecting internal systems to third-party platforms, each one designed as a reusable pattern rather than a one-off — that is the difference between the second integration taking days and taking weeks.

Design

The hero carries the entire product thesis in two buttons, Blur My Card and Play Rate-To-Reveal: grading as a game you play rather than a report you receive. The registry navigation sitting behind them does the credibility work.

Development

SvelteKit and TypeScript throughout, released to production weekly with zero downtime on two-week Agile iterations, with automated QA and data validation folded into the delivery pipeline rather than bolted on at the end.

Integrations & APIs

AI scoring services, third-party collectibles platforms, and the workflow pipelines carrying data between them and internal systems.

Outcomes

• Lighthouse 95+ with sub-second load times in production
• Zero-downtime weekly releases on two-week Agile iterations
• 30% less manual development effort from automated QA and data validation in the pipeline

Discovery

Two jobs running in parallel. Client-facing: performance, accessibility, and engagement audits across major brands including United Healthcare — healthcare is where accessibility stops being a checklist item and starts being the product. Internal: an inventory of the agency’s AEM and WordPress estate to find where properties had drifted apart and what that drift was costing.

Architecture

I served as the architecture authority across both ecosystems — governance standards, design patterns, component libraries, and the performance and security benchmarks every property was held to. Architecture options, risk assessments, and roadmap tradeoffs went to executive stakeholders to drive platform investment decisions. That work is invisible when it goes well, which is the point.

Development

WordPress systems refactored for speed and engagement, developer workflows and release process streamlined across teams, and AI-driven analytics wired into campaign reporting.

Integrations & APIs

The NetSuite–AEM ERP integration unified billing, content, and financial reporting data across on-premise and cloud systems. EDI and CRM integrations followed on the same backbone, and Snowflake became the governed data-warehouse reporting layer for cross-platform campaign and financial data. Both replaced processes that had previously run on spreadsheets and trust.

Outcomes

• AI-driven analytics integration — campaign precision up 15%
• WordPress systems refactored for 25% more speed and engagement
• Developer workflows streamlined for 35% better efficiency
• API-driven data pipelines cutting manual processes by 40%

Discovery

I led platform discovery and requirements sessions with business and technical stakeholders, producing technical context diagrams and as-is/to-be architecture to align people before the build rather than during it. Six years in, the useful discovery stopped being “what should this feature do” and became “why do four of our properties solve this differently.”

Architecture

I established the architectural, security, and performance benchmarks that were then adopted across all client engagements, and diagnosed the WordPress platform issues underneath the recurring symptoms — downtime and load time were the visible part of a structural problem.

Design

A bespoke gallery experience built with the design and marketing stakeholders, for a catalogue where the photography is the product, plus product displays streamlined around what analytics said people actually did rather than what the room assumed.

Development

A custom WooCommerce module, a React interface layer, and the Storybook component library the whole team worked from — which also let stakeholders sign off components before production. Alongside the build I supported multiple concurrent Agile delivery teams through sprint planning, code review, and enforcement of the quality metrics, and ran multi-channel A/B testing across email and landing pages.

Outcomes

• 40% increase in online sales efficiency from the custom WooCommerce module
• 20% lift in completed CTAs from streamlined product displays
• 35% better engagement and 20% lower bounce, guided by Google Analytics
• 30% shorter deployment cycles from the Storybook library, plus 30% less downtime after the CMS remediation

Discovery

Adobe Experience Manager and Magento are both entirely convinced they own the customer experience. The engagement opened by mapping which system should own what — content in AEM, inventory and cart in Magento — and where the seam between them would become visible to someone actually shopping.

Architecture

Custom plugins bridging the two platforms, so a single purchase journey crosses systems without the handoff showing. Designed so content editors kept working in AEM and merchandisers kept working in Magento, neither aware of the other.

Design

Custom AEM components built against the existing wireframes, and a rework of the subscription configuration UI — the toner-colour selector is the front door to the business model, and it had to make a large configuration matrix feel like a dropdown.

Development

I rebuilt the Magento 2.4 subscription module to offer custom toner colour options, which sounds small and was not: toner subscriptions are the business model, and the configuration matrix behind that dropdown is where the complexity lives. CMS components were refactored and performance-tuned for load time, and automated QA regression coverage went in behind the release.

Integrations & APIs

The AEM–Magento integration itself, plus REST calls inside React components fetching real-time printer information, with state management driving automatic content updates on the page.

Outcomes

• 30% increase in user engagement and a 15% higher conversion rate from the AEM–Magento integration
• 35% faster page loads from refactored CMS components, feeding a 20% gain in satisfaction scores
• 20% higher satisfaction and 10% more subscription sales from the module rebuild
• 20% less manual testing overhead from automated QA regression coverage

Discovery

A legacy estate that had grown by accumulation: on-premise CMS platforms, several separate payment processors, and POS, order, and product data that did not agree with each other. The current-state assessment came before the migration plan, because the migration was never the actual problem.

Architecture

I designed and spearheaded the transition to Magento Enterprise 2.4, and the move of legacy on-premise CMS platforms onto cloud-hosted WordPress, Magento, and AEM — including stakeholder coordination and timeline management. That was the half the business could see. I also designed the workflow components that replaced several legacy systems outright.

Development

Java solutions against SQL Server, personalization and segmentation logic in Salesforce Commerce Cloud, and SQL Server performance tuning that cut query response times and improved reporting accuracy.

Integrations & APIs

The harder half. WebMethods and ERP integrations in Java and SOAP/XML unified POS, order, and product data across commerce and content systems, carrying EDI, CRM, and data-warehouse exchanges on the same backbone. POS and transaction data fed Snowflake to drive segmentation modelling in Salesforce Commerce Cloud — better targeting precision, less audience overlap between campaigns that had previously been competing for the same person. Marketo and Litmus were integrated with the AEM-driven properties, and the scattered legacy payment systems were consolidated onto a single Braintree integration.

Outcomes

• Conversions up 18% and cart abandonment down 15% via AEM personalization and targeted campaigns
• 30% less manual financial reconciliation from the WebMethods and ERP integration
• Send-time optimization and A/B testing through Marketo and Litmus
• Legacy payment systems consolidated onto a single Braintree integration

Discovery

What I inherited: a legacy ecommerce platform, a 10,000-product catalogue, half a million customer records, a tangle of separate payment processors, and a five-person team deploying without version control. I strategized the transition before touching any of it, because the sequence mattered more than the target platform.

Architecture

WordPress for content and Magento 2.3 for commerce, with custom store views for warehouse sourcing so the catalogue reflected where stock actually was rather than where the head office believed it was.

Design

Front-end and template work across both platforms as part of the migration, rather than a straight port of the legacy presentation.

Development

The migration itself: 10,000+ custom and simple products and 500,000 customer records carried across. A dropped row in that import is not a statistic, it is a long-standing customer who cannot log in and does not call to tell you — they just leave. I also put Git across every environment in a shop that had been deploying without it, and did the performance and security pass on the WordPress side.

Integrations & APIs

The legacy payment processors consolidated onto a single Braintree integration.

Team

The part I am most pleased with is the least technical. I mentored a five-person development team through PHP, WordPress, and component architecture, then wrote the standards down — code review criteria and documentation templates — so the improvement outlasted the engagement.

Outcomes

• 20% improvement in site performance and reliability, and a lift in ecommerce conversion
• 30% higher team productivity and code quality after mentoring
• 40% faster onboarding for new developers from code review standards and documentation templates

Discovery

The site’s actual job is to move a worried person from “my doctor mentioned this” to a booked appointment. Framing it that way changed the priority order: the physician locator became the most important thing on the site, and the email program became the thing that closes the gap for everyone not ready on the first visit.

Development

Customized HubSpot landing pages and modules built for campaign reuse rather than one-offs, plus direct and marketing email templates written to survive the rendering quirks of every client a fifty-year-old patient might plausibly be reading in.

Integrations & APIs

I built the physician locator as an API-driven mapping application rather than a static list, so it stayed correct as the provider network changed — a static list of cardiologists is wrong within a quarter and nobody notices until a patient does.

Outcomes

• 30% boost in lead conversions from the custom HubSpot landing pages and modules
• 23% longer visit duration from the API-driven physician locator
• 38% lift in opens and click-through from the redesigned email templates

Discovery

I led the team meetings driving WordPress platform enhancements — the roadmap came out of those sessions rather than off a ticket queue — and contributed to the technical specifications and architecture design that came out of them.

Design

Advised on site design and UX/UI enhancements across the marketplace, working with the design team rather than around them.

Development

Custom PHP-based WordPress themes and platform enhancements built against those specifications, aligned to business goals rather than to a feature list, plus the technical SEO program that turned out to be the highest-return work of the engagement.

Delivery

The deployment side mattered as much as the features. I deployed auto-build procedures to production with real testing and validation, replacing a release process that worked fine right up until the person who knew it took a holiday.

Outcomes

• 35% increase in conversions from the SEO program
• 25% more site traffic from platform enhancements
• Automated build and deploy to production with rigorous testing and validation

Discovery

A genuinely devoted customer base raises the stakes on the product page — these buyers know the seatbelt buckle by name and will notice if a spec is wrong. The other finding was in the analytics rather than the brief: 2014 was the year mobile stopped being optional, and bounce on small screens was the largest single number on the table.

Design

Custom product detail, category, and landing page layouts built around the specification detail those buyers came for, and a responsive rebuild across the store.

Development

The Magento template work behind those layouts, plus custom JavaScript and CSS written specifically to get page weight down rather than to add anything.

Security & compliance

I led the PCI self-assessment questionnaire and scan through to full compliance — unglamorous, non-negotiable, and the kind of work that only becomes visible when it has not been done.

Outcomes

• Add-to-cart up 20% and retention up 40% from the custom PDP, category, and landing page layouts
• 30% faster page loads from custom JavaScript and CSS
• 25% lower bounce from the responsive build
• PCI SAQ and scan led through to full compliance

Discovery

Campaign platforms have an unusual failure profile: a successful launch is indistinguishable from a denial-of-service attack. Scoping every build started with the ceiling rather than the feature list — what happens on the day the campaign works.

Architecture

Collection, moderation, and playback designed to hold at campaign scale, with the moderation queue sitting between the public feed and the page so a brand hub could never publish something nobody had looked at.

Development

The custom UGC platforms and the video tooling underneath them. The rebuild I remember is the video player: buffering is the metric nobody puts in the brief and everybody notices, and halving it did more for engagement than any feature shipped that quarter.

Integrations & APIs

Social and content platform APIs feeding the hubs, normalized so one moderation pipeline served every source.

Outcomes

• Custom UGC platforms and video tooling shipped for Fortune 500 brands
• Video player rebuild cutting buffering 50%

Discovery

Stock Magento in 2012 was slow, and the industry consensus was that this was simply the price of Magento. Benchmarking said otherwise. The entire engagement was built on that finding, which is the only reason the number at the end is what it is.

Architecture

Front and back end rebuilt around performance as a constraint rather than tuned for it afterwards — the two produce very different systems and only one of them holds.

Design

Storefront and product presentation designed around motion, because in a dancewear catalogue a still photograph of a garment is missing the thing being sold.

Development

The Magento build front and back, plus a custom HTML5 video player. That component dates the project precisely: Flash was dying but not yet dead, and building native video in 2012 meant handling every browser’s idea of a codec yourself.

Integrations & APIs

Analytics and conversion instrumentation across the full funnel, so the optimization work had numbers behind it rather than instinct.

Outcomes

• 70% performance gain over stock Magento
• Custom HTML5 video player, built ahead of broad browser consensus
• Analytics and conversion instrumentation across the full funnel

Discovery

The brand had to hold up in three places at once: a website, printed collateral handed out at the events themselves, and the email program that filled them. Three media that punish a logo in three different ways, and the mark had to be resolved against all of them before anything got built.

Design

I directed the creative — full identity across site, print, and email, including the event collateral.

Architecture & development

Then I built it: the site, the CMS front and back end, and the SEO program. Being on the hook for the build is a useful constraint on the art direction — you stop specifying things you know you will resent implementing.

Outcomes

• Full identity across site, print, and email
• CMS front and back end, built to be run by the event team
• SEO program, and an 80% performance gain over standard builds of the period

Discovery

The brief was a Flash-based game that needed to run on phones, which in 2011 meant it needed to stop being Flash. Apple had published Thoughts on Flash about a year earlier and most agencies were still arguing about it; here the client’s own analytics had already settled the argument, so the discovery work was mostly showing people the numbers.

Development

Ported the game to HTML5 and jQuery Mobile, halving load times in the process, then built the companion iOS and Android applications on top of the same game logic.

Outcomes

• Flash game ported to HTML5 and jQuery Mobile, halving load times
• Companion iOS and Android applications shipped

Discovery

A catalogue where the photography carries the sale and the specification table closes it — and, underneath that, product content the search engines could not reach at all. The structural finding was worth more than anything on the feature list.

Design

UI design across the catalogue, category, and product pages, built around large photography without letting it push the specifications off the screen.

Development

Production CMS build against that design, plus the commerce integration behind it.

SEO

The 220% visibility lift is the number that still surprises people, and it is only possible on a site that started with a structural problem. Most of the win was making the catalogue legible to crawlers rather than anything clever.

Outcomes

• UI design through production CMS build
• Commerce integration
• SEO program lifting search visibility 220%

Discovery

A developer with forty years of Manhattan buildings behind it and nowhere to show them. The site had two jobs: sell new developments that mostly existed as renderings, and make the back catalogue, from the Excelsior in 1965 to the W Times Square, read as one body of work.

Design

A set of competing concepts — Classic, Clean Lines, Color Impact — before one comp was approved. The client’s March notes asked for restraint: the cube-wipe transitions became fades, the type came down a point, and Development History became a timeline, each building placed on it by decade.

Development

Flash 8 and ActionScript 2, with Tweener for the motion. Deep linking ran through SWFAddress, so every section had its own URL, the back button worked, and Google Analytics could count pages — none of which Flash did on its own. The stage scales to the window, with bitmap smoothing so the architectural photography held up at any size.

Outcomes

• Launched April 2009
• Deep-linked, analytics-tracked Flash site
• Development history built as an interactive timeline
• Still running today from the original SWF

Discovery

Every campaign started as a data question rather than a design one: what fields exist per recipient, which of them can be trusted, and what the page does when one is missing. Personalization at volume fails on the empty field, not the full one.

Architecture

Everything was built against a proprietary CMS on Linux, Apache, and Tomcat. Six years working inside somebody else’s CMS teaches you the thing no framework tutorial does: what a platform’s assumptions cost you three years later, and how to design so the answer is “less than it could have been.”

Development

Web and email builds in JSP against that CMS, personalized per recipient at campaign volume — years before “marketing automation” was a category with a Gartner quadrant.

Outcomes

• Six years of web and email builds against a proprietary CMS
• Linux / Apache / Tomcat production environment
• Variable-data personalization across web and print at enterprise volume

Discovery

Eight years of running my own shop meant every engagement started with me in the room for the sales conversation, the requirements, and the estimate. That is where the habit of assessing current state before proposing anything came from — when you own the estimate, an optimistic assumption is expensive and personal.

Architecture & design

Architecture, build, and creative direction were not separate offerings so much as the same job seen from different ends. A small practice does not get to specialize, which turns out to be the most useful training available.

Development

Front-end and CMS builds across the client roster, delivered solo or with contractors I directed. Nearly everything from this period is gone from the live web; the capture beside this is the later portfolio that carried the domain forward.

Outcomes

• Independent practice sustained eight years across enterprise, SMB, and agency clients
• Architecture, build, and creative direction owned end to end on every engagement

Discovery

Site strategy and information architecture, done as the first phase of every project rather than as a deliverable sold separately. The earliest work in this archive and the point where the sequence — strategy, then structure, then surface — became habit.

Design

Art direction across screen and print on the same projects, so the collateral and the site were one thing rather than two interpretations of a brand.

Development

Cross-platform build, which in 2003 meant Netscape as well as Internet Explorer. A reasonable illustration of how much the job has changed: the layout problems of that period were solved with tables and spacer GIFs because there was nothing else.

Outcomes

• Site strategy and information architecture
• Cross-platform build across the browser matrix of the period
• Collateral production alongside the digital work

Discovery

The useful question was not what the store needed on launch day but what it would need after growth — how the catalogue would be cut two seasons later, and which of today’s conveniences would be load-bearing by then.

Architecture

Catalog structure and checkout flow designed for that later business rather than the current one. It is the difference between a store that scales and one that gets rebuilt in eighteen months by someone cursing your name.

Development

The Magento build against that architecture, front and back.

Integrations & APIs

Integration work tying the storefront to the systems behind it, each designed as a reusable pattern, so inventory, orders, and product data stayed consistent without anyone reconciling spreadsheets on a Friday afternoon.

Outcomes

• Catalog and checkout architecture built with growth headroom
• Third-party integrations designed as reusable patterns

Discovery

Music sites are media-heavy by nature, and the usual outcome is a beautiful page that takes eight seconds to become useful, by which point the visitor has gone back to the streaming service. That constraint was established before the design started rather than discovered after launch, which is why it was survivable.

Design

Identity and visual design end to end, built to carry artwork at scale without the page becoming a gallery the visitor has to wait for.

Architecture & development

The front end was tuned around first render from the outset rather than optimized after the fact — the only version of performance work that actually holds. A CMS behind it so the roster could be updated without me.

Outcomes

• Full identity through production build
• Media-heavy pages tuned for fast first render
• CMS handover for independent roster updates

Discovery

Hospitality has a context problem that most responsive briefs quietly skip: the person reading your menu is holding a drink, the room is dark, and they have about fifteen seconds of patience. Everything downstream was scoped against that situation rather than against a desktop comp.

Design

Booking and menu flows designed mobile-first for that context and desktop second. Design, architecture, and build were all mine, which meant the responsive behaviour was a design decision made up front rather than a development compromise negotiated at the end.

Development

WordPress build with the storefront integrated into the brand site, so ordering and reservations did not hand the visitor off to somebody else’s checkout mid-decision.

Outcomes

• Mobile-first booking and menu flows
• Storefront integrated with the brand site

Discovery

One constraint shaped every decision here: the people maintaining this site after I left are volunteers, not developers. That rules out most of the interesting answers, and it is a genuinely better brief than it sounds — it forces you to justify complexity to somebody who will never read the documentation.

Architecture

Restructured around what a volunteer can safely edit. Anything clever enough to need me later was the wrong answer, so the clever parts were removed rather than documented.

Development

An accessibility pass across templates and content, and security hardening on a WordPress install that had accumulated the usual decade of plugin debt.

Outcomes

• Restructured for volunteer maintenance
• Accessibility pass and security hardening