I've spent 15+ years on frontend, platform, and full stack teams in enterprise environments. I'm good at stabilizing teams that have lost their way and getting them shipping again, and I care about the engineers on those teams growing while we do it.
Where I focus and how I work.
Most of my recent work has been coordinating distributed systems, APIs, and data pipelines across large organizations. I lean on architecture, experimentation, and operational discipline rather than late-night heroics, and I push my teams to do the same.
I run as much as a product manager as an engineering manager. As a certified Professional Scrum Master (PSM I), I support agile practices to enhance team productivity and project outcomes, and I align technical execution with the organization's goals so the team is empowered to ship work that actually moves the business. In practice that means owning the roadmap, decomposing requirements, prioritizing the backlog, writing acceptance criteria, and aligning Product, Design, Data, and Engineering before a sprint kicks off.
Before management I spent close to a decade as a senior frontend and full stack engineer at retail, marketing, SaaS, and DoD companies. That background still shapes how I run delivery today. I push for small ships and honest measurement, and I'd rather fix the system that produced a bug than the bug itself.
Outside work I'm a hands-on AI engineer. I design and ship production LLM applications on the Anthropic API — RAG, prompt caching, structured-output streaming, multimodal vision features, LLM-as-judge evaluation patterns — and run them on self-hosted infrastructure. I work alongside open-source ML when the task calls for it: Demucs for audio source separation, RVC and so-vits-svc for generative-audio fine-tuning, training scaffolds and eval harnesses to make the pipeline reproducible. The side work is how I stay current on the frontier and how I learn new stacks before I'm making decisions about them on a real team.
Production LLM applications and ML pipelines I've designed, shipped, and operate end-to-end. Anthropic Claude on the LLM side; open-source models (Demucs, RVC, so-vits-svc) on the audio-ML side; Next.js, Node, and TypeScript across the application layer. All self-hosted on a home NAS behind Cloudflare Tunnel or deployed independently. Swipe or scroll through the showcase — click any tile to open the live tool.
Conversational AI built on the Anthropic Messages API. Persona document plus a curated career corpus served as a single prompt-cached system prefix — first request pays the cache write, every subsequent request hits the cache. End-to-end SSE streaming, system-prompt design, and per-request addendum injection (kept out of the cached prefix) for tailored content surfaces. Designed defensively against prompt-injection extraction.
Production SaaS rack planner for audio engineers with a multimodal AI feature: upload a photo of an existing rack and a vision-language model returns a structured component inventory the user can edit. Demonstrates production VLM integration, image-to-structured-output prompting, and graceful UX for uncertain model output. Next.js 15 / React 19 / TypeScript / Tailwind v4 / Supabase, 532 tests.
An evaluator that scores creative work against a 34-item rubric and returns per-criterion verdicts with line-level citations — an LLM-as-judge / eval-harness pattern applied to lyrics. Demonstrates structured-output prompt design, rubric scoring, and citation-style attribution. Zero-dependency Node 20 server (~250 LOC), Dockerized for self-hosted deployment.
Audio-ML pipeline that extracts spectral, stereo-image, and dynamics features from a reference track, then summarizes them via an LLM into a structured analytical breakdown — tonal balance, stereo image, dynamics signature, arrangement landmarks — for downstream mix and master decisions. Demonstrates pairing classical audio feature extraction with LLM-driven structured output. Self-hosted on the NAS behind Cloudflare Tunnel.
A working portfolio of prompt-engineering case studies: each toy explores a different LLM interaction pattern beyond the chatbot — generative recommendation, hallucination-as-feature, comparative ranking, persona-driven critique, latent-space navigation. Useful as a sandbox for testing prompt structures, output shapes, and product surfaces around LLMs. Self-hosted Node + Anthropic API.
Spatial recommendation surface: drag a point across two semantic axes (year, energy) and the audio crossfades to whatever lives at that coordinate. Demonstrates real-time LLM-driven ranking, embedding-style spatial interaction, and progressive UI design for AI features. Promoted out of the Music Toys collection because this interaction model proved the strongest.
End-to-end generative-audio pipeline: Demucs source separation → vocal stem extraction → dataset curation → RVC / so-vits-svc fine-tuning → audition harness for subjective and objective model evaluation. Demonstrates audio-ML data engineering, fine-tuning open-source generative speech models, and building eval scaffolding for generative-audio research.
Customer-facing configurator with live preview, Stripe deposit capture, Resend transactional email, and an admin operations surface — shipped solo end-to-end using AI-assisted coding (Claude Code) as the primary development workflow. A working example of human-in-the-loop AI-assisted delivery alongside the AI products themselves. Next.js 16 / Tailwind v4 / Zustand / Supabase / Stripe / Resend.
Fifteen years of client and platform work, plus the product I build and run myself. Click any tile to open the full case image — or the film tile to watch the walkthrough.
A walkthrough of the application I designed, built, and operate solo — gear search across a 350+ component library, drag-and-drop rack layout, automatic signal-flow wiring, and the generated tour documentation engineers hand to a venue. Every surface in the film is production software: Next.js 15 / React 19 / TypeScript / Tailwind v4 / Supabase, with live subscription billing and an automated SEO pipeline behind it.
Architected a body of design-strategy work across AE's commerce ecosystem — a unified PDP recommendation experience, swipeable PLP image onboarding, a centralized cross-platform recommendation framework, and a system-level product-badging architecture. Each initiative applies platform-thinking and design-systems maturity to high-traffic commerce surfaces, with a confidence-first lens on customer decision-making.
Developed a trio of high-performance recommenders for Aerie’s product detail pages, utilizing Machine Learning to generate high-affinity candidates. This system features a sophisticated experimentation layer that dynamically determines placement and ranking logic, optimizing for maximum user engagement and conversion.
Engineered a modular recommendation ecosystem to transform single-item browsing into a curated styling journey. By leveraging real-time inventory and deep attribute-based filtering, these engines streamline the path to purchase and drive significant growth in average order value across mobile and web platforms.
Architected a specialized "Mix & Match" engine designed to empower users with personalized set-building capabilities. Through advanced item-to-item mapping and automated candidate generation, this tool simplifies the selection of coordinated separates, effectively bridging the gap between discovery and transaction.
Architected an internal component library for Carnegie Mellon's Software Engineering Institute. Codified the institute's visual language into reusable UI primitives and downloadable kits, giving engineers across the org a shared foundation for shipping consistent, on-brand interfaces.
Built the Storybook surface for the SEI Design System — searchable, interactive component documentation that lets engineers explore behavior, states, and usage patterns without reading source. The discovery layer that turned the component library into a self-serve resource.
Engineered an internal workflow tool to replace a brittle email-based approval process. Routing logic, real-time status tracking, and per-user inboxes for pending and overdue items consolidated cross-team approvals into a single auditable system.
Designed a queryable staff records and reporting platform to replace a sprawl of disconnected spreadsheets. Tied directly into the institute's identity provider, surfacing clean, authoritative people data for org-wide reporting.
Built a search interface over Oracle's chargeback codes for SEI internal billing — collapsing a multi-minute spreadsheet hunt into a single typed query. Small surface, big multiplier on the team's daily friction.
Native iOS shopping experience for Burberry covering browse, product detail, and checkout. Tech lead on architecture and delivery quality, ensuring the app held the brand's standard while shipping at retail-cycle pace.
Native iOS and Android shopping apps for Levi Strauss, built as a coordinated platform engagement. Architectural patterns established here became the template for subsequent retail clients in the agency's portfolio.
Rider ticketing experience for the Las Vegas Monorail — a React web purchase flow paired with a React Native mobile app over a shared Redux and TypeScript layer. Beacon presence detection at stations, in-app purchase, and a gate-scannable barcode at the turnstile. End-to-end mobile-first transit access in a single tap.
Account management, ride history, multi-ticket bundle purchasing, and the barcode-display turnstile flow — the supporting surfaces that wrap the core ticketing experience and turn casual riders into regulars.
The operational backbone behind the Monorail's customer-facing apps. Fare structures, validity windows, and staff tooling — the engine that kept turnstile transactions consistent across every channel, containerized with Docker for consistent deploys.
Operations dashboards surfacing ridership and ticketing data for the Monorail's leadership team. Built atop the same platform as the customer apps for unified data flow from turnstile to executive review.
End-to-end redesign of Trinds.com — brand refresh, restructured content architecture, and a hand-built marketing site backed by a lightweight CMS. A faster, sharper expression of the brand.
Developed a comprehensive visual identity and brand system for a boutique Pittsburgh confectionery. This included establishing the typography, color palette, and packaging direction, alongside a custom storefront site designed to translate the brand's premium aesthetic into a digital retail experience.
Architected a 3D maze puzzler from the ground up, serving as the sole designer and engineer. I managed the end-to-end production pipeline—including game mechanics, art direction, and frontend implementation—to ship a polished experience within a rigorous, self-imposed development cycle.
A multi-disciplinary showcase of game development excellence—specializing in high-poly character pipelines, comprehensive UI/UX, and intelligent, reactive camera systems that breathe life into every player action.
The last fifteen years in summary. Full detail is on the resume.
Engineering Manager for Product Discoverability UI, working with ML personalization, recommendations, and UGC to improve core PDP and PLP experiences. Inherited a team that hadn't shipped in multiple sprints and turned it into a group consistently delivering 20+ features across a $5B+ ecommerce platform. Led delivery of the ML-powered feature set — the Aerie recommendation engine, a multi-channel personalization suite, a “Mix & Match” dynamic bundle engine, and generative AI integration tools. Led an international team of 12 engineers (FTEs and contractors), delivered 2.56% Revenue Per Visitor and 1.7% lift in unique conversions on Activewear, improved team velocity by 20% while reducing production bugs from 65 to 44 (32% reduction), and shipped 116 of 595 total deployments to production (19% of org output) with zero critical incidents. Operated as a hybrid Engineering Manager and Product Manager — owned the roadmap, requirements, prioritization, and stakeholder alignment alongside delivery, and ran the agile practice (sprint planning, refinement, retros, demos) as a certified Professional Scrum Master (PSM I). Owned the end-to-end A/B and CRO program on Optimizely — experiment design, audience segmentation, feature-flag rollouts, experiment QA, and weekly stakeholder reporting that fed back into product prioritization, paired with Quantum Metric session insights for customer-behavior analysis.
Technical lead for enterprise web platforms, building large applications in Vue, Nuxt, and TypeScript. Built the SEI Design System, an internal frontend component library in Tailwind and SCSS that got picked up across the organization, documented in an interactive Storybook surface so teams could discover components on their own. Replaced a brittle email-based approval process with an automated workflow engine handling routing logic and status tracking, built a centralized identity portal giving queryable staff records in place of scattered spreadsheets, and built a financial data query interface over Oracle chargeback codes that collapsed a multi-minute spreadsheet hunt into a single typed query. Integrated UI components into legacy systems as part of a broader platform modernization effort, consuming REST APIs backed by Node.js, Python, and MySQL services. Kept releases reliable through Selenium coverage, Bamboo CI/CD, and Docker builds. Led architecture reviews, mentored frontend engineers, ran product roadmapping and usability testing with stakeholders, and set patterns the team could actually maintain long term.
Technical lead on commerce platform work for major retail brands, building React and Redux applications in TypeScript and owning implementation quality, architecture decisions, and delivery timelines across mobile/web and consumer systems with real traffic. Tech lead on client engagements for Burberry — a native iOS shopping app covering browse, product detail, and checkout — and contributed to Levi Strauss’s cross-platform iOS and Android retail apps, working in Swift and Objective C alongside the React web clients against shared REST APIs and Node.js services. Contributed to the shared design system and production design assets in Photoshop and Illustrator. Mentored junior and mid level engineers through code reviews, pairing sessions, and architecture guidance, and sat between client stakeholders and engineering teams, making sure business requirements turned into solutions that could actually scale.
Independent contract engagement, concurrent with full-time roles, delivering the complete rider ticketing suite: a React web purchase flow paired with a React Native mobile app over a shared Redux and TypeScript layer. Built the rider lifecycle end to end — account management, ride history, multi-ticket bundle purchasing, and the barcode-display turnstile flow — over an operational backbone of fare structures, validity windows, and staff tooling, plus ridership dashboards for the Monorail’s leadership team. Containerized with Docker for consistent deploys.
Full stack engineer on an enterprise content marketing SaaS platform, working across React, Angular, and Redux on the frontend and Java and Kotlin on the backend, all in a Spring architecture. Rotated across Circle teams (Rapid Response, Core, Integrations). Rapid Response was mostly backend Java, diagnosing and hotfixing issues customers reported, fast turnaround. Designed and consumed REST APIs against PostgreSQL and Node.js services, with Python tooling for data and automation tasks. Prototyped and designed UI/UX patterns for the platform’s shared pattern library, which grew into a design system and brought consistency across the product. Built a WYSIWYG editor integration so content creators could produce and manage rich text content without leaving the platform. Led scrum conversations, reported team metrics, and kept build automation and code review running on Jenkins, Bamboo, and Gradle.
Built native iOS and web applications for clients including Rite Aid, Cooper Tire, PA Lottery, Make A Wish Foundation, Payless, and True Value. Also built an iOS asset management app using Bluetooth beacons to track employee locations across facilities in real time.
Frontend developer and primary technical decision-maker during live sessions, building CMS-driven platforms across multiple brands and campaigns. Handled wireframing and UI/UX design alongside implementation, building on PHP and MySQL with JavaScript frontends.
Transitioned from graphic design into full stack web development on a LAMP stack (PHP, MySQL, JavaScript). Technical owner for the biggest client accounts, delivering campaigns across web, social, and advertising channels. Owned wireframing and UI/UX design end to end, producing campaign and interface assets in Photoshop and Illustrator.
Click either page to open the PDF, or download the file. There is also a version focused on the AI and LLM engineering work.
What I make outside of work.
I write and record my own music — alternative rock, made at home here in Pittsburgh. I play the parts, mix it, and put it out myself. It lives on its own site.