Selected Write-ups From Recent Work

Overview

Dealertrack is a leading automotive retail platform used by thousands of dealerships nationwide. This project centered on integrating a third-party fraud analytics solution into the existing Finance & Insurance (F&I) workflow — a high-stakes environment where deal accuracy, compliance, and speed all carry significant weight.

The goal was to make fraud detection feel native to the dealer's existing process, not bolted on.

The Problem

F&I managers were navigating compliance checks manually — a fragmented, time-consuming process that created friction at a critical point in the deal flow. Without real-time fraud signals surfaced in context, dealers were left making decisions with incomplete information, and compliance became reactive rather than proactive.

At the start of this engagement, there was no prior UX research on these workflows and significant technical constraints to design around. Everything had to be built from the ground up.

My Process

Heuristic Audit. I began with a thorough heuristic audit of the existing Dealertrack F&I interface, mapping pain points across the current workflow. This gave me a clear baseline — where users were losing time, where the interface created ambiguity, and where compliance tasks were creating the most friction.

User Flows & Wireframes. From the audit findings, I mapped revised user flows that repositioned fraud detection as a natural step in the deal process rather than an interruption. Wireframes explored how to surface verification states, alerts, and document handling without overwhelming F&I managers mid-deal.

Prototyping & Usability Testing. I built clickable prototypes that were used directly in usability testing sessions. Testing focused on key interaction moments: understanding verified vs. non-verified states at a glance, navigating a mobile-first identity verification experience (license scan, selfie capture), and interpreting fraud alerts clearly enough to act on them with confidence.

High-Fidelity Design. Final designs included custom components and a modular widget framework designed to accommodate future fraud detection layers — built for this MVP, but architected to scale.

Key Design Decisions

Clarity of state, always. One of the core challenges was communicating verification outcomes — verified, pending, flagged — in a way that was immediately readable under pressure. I designed distinct UI states with clear visual hierarchy so dealers could orient themselves at a glance without parsing dense compliance language.

Mobile-first identity verification. The identity capture flow (license scan + selfie) had to feel fast and trustworthy without stalling the sales process. I designed this as a streamlined, step-by-step experience that could be completed on a mobile device with minimal instruction — intuitive enough that it didn't require training.

Turning compliance into a guide, not a gatekeeper. Rather than surfacing alerts as blockers, the design framed fraud signals as decision support — guiding dealers toward faster, safer deal closures. Messaging was written to be directive without being alarmist.

Designing for auditability. Document handling was designed with both the dealer and future compliance reviews in mind — clear, structured, and traceable.

Outcome

• 16% reduction in time-on-task across key workflow touchpoints
• 12% improvement in task completion rate measured through usability testing
• Delivered an MVP that validated technical feasibility of the integration within Dealertrack
• Established a modular design framework ready to support future fraud detection enhancements

Reflection

Working without existing UX research and within tight technical constraints pushed me to rely heavily on first-principles thinking — the heuristic audit became the foundation everything else was built on. The biggest design challenge wasn't the complexity of fraud detection; it was making something inherently compliance-heavy feel like a natural, low-friction part of a dealer's day.

Overview

A New York City electrical contracting company had no web presence, no online marketing strategy, and no way for potential customers to find or evaluate them digitally. This project involved designing and launching a production-ready brochure website from the ground up — on an accelerated timeline — using an AI-assisted workflow that maintained the highest standards of UX, visual design, accessibility, security, and search optimization.

The site is live and actively generating leads.

The Problem

Without a website, the business was invisible online at a time when even trade and contracting clients vet vendors digitally before making contact. There was no brand presence to speak of, no way to capture inbound leads, and no foundation for future digital marketing efforts.

The challenge wasn't just building a website — it was building the right website: one that reflected a credible, professional brand, performed well in search, and converted visitors into real inquiries.

My Process

Brand Grounding. The project started with the client's branding guidelines, which I used as the design source of truth throughout. Rather than interpreting brand standards loosely, I built a structured design.md file — a detailed design specification I authored that codified front-end design decisions: typography, color application, spacing, component behavior, and visual tone. This file became the bridge between brand intent and built output.

AI-Assisted Development with Claude Code. Using Claude Code, I translated the design.md specification into a fully realized front-end. This wasn't a prompt-and-hope workflow — the design spec I wrote gave the AI a precise, structured brief to execute against. I used Puppeteer to take and analyze screenshots at key breakpoints, allowing me to visually QA the output, catch inconsistencies, and iterate rapidly. The result was a site that looked and felt like it was designed by hand, because the design direction was.

UX Best Practices & Accessibility. Throughout design and development, the site was held to established UX standards and W3C accessibility guidelines. This included semantic HTML structure, sufficient color contrast ratios, keyboard navigability, descriptive alt text, and accessible form labeling — ensuring the site works for all users regardless of ability or assistive technology. Accessibility wasn't treated as a checklist item at the end; it was built into the design specification from the start.

Security & Compliance. Web forms were reviewed and hardened against common vulnerabilities. All phone number instances on the site were implemented in a way that prevents automated spam harvesting — a practical consideration for a local trade business that relies on phone inquiries as a primary lead channel.

Search Optimization: SEO, GEO & AEO. The site was optimized across three dimensions. Traditional SEO ensured strong crawlability and keyword relevance for NYC electrical contracting searches. GEO (Generative Engine Optimization) prepared the site's content for AI-driven search surfaces that increasingly surface direct answers rather than blue links. AEO (Answer Engine Optimization) structured content so that voice and AI assistants can accurately represent the business in response to relevant queries. This layered approach future-proofs the site's discoverability as search behavior continues to evolve.

Key Design Decisions

Design spec first, build second. The design.md file was the most important artifact of this project. By writing a thorough front-end design specification before any code was generated, I maintained creative control over the output and ensured the final site reflected deliberate design decisions — not AI defaults. The spec is reusable and scalable for future site updates.

Designing for trust in a trade context. Electrical contracting clients are evaluating credibility, professionalism, and local expertise in seconds. Every design decision — from typography weight to how services were structured and described — was made with that trust-building moment in mind.

Performance and protection as UX. Security compliance, spam prevention, and accessibility aren't typically framed as UX decisions, but they are. A compromised form, a harvested phone number, or a site that excludes users with disabilities all degrade the experience for real people. These were treated as design requirements, not afterthoughts.

Outcome

• Site is live and actively generating inbound leads for the client
• Full brochure website delivered on an accelerated timeline using an AI-assisted workflow
• Passed security compliance checks across all web forms
• Built to W3C accessibility standards, ensuring an inclusive experience for all users
• Optimized for SEO, GEO, and AEO — positioning the client for discoverability across traditional and AI-driven search surfaces
• Demonstrated that production-quality digital products built with AI can meet the highest standards of UX, design craft, and accessibility when guided by a skilled designer

Reflection

This project is proof of concept for a new way of working. AI didn't replace the design process — it executed it. The quality of the output was a direct function of the quality of the brief: a structured design specification, a clear brand reference, a commitment to accessibility from day one, and a systematic QA process using Puppeteer to catch and correct what needed refinement. The result is a live product generating real business results for a client who had nothing before.

As AI tools become more capable, the designer's role shifts toward authorship and direction — knowing what to specify, what to scrutinize, and what good looks like. This project is a demonstration of that skillset in practice.

Overview

Dealertrack's F&I platform powers the deal-making process for thousands of automotive dealerships. As deals move through the pipeline, they generate a significant volume of documents — contracts, compliance forms, lender paperwork, customer agreements, and more. This project involved designing the Digital Deal Jacket: a centralized document management system built from scratch within the Dealertrack and Deal 360 ecosystem.

There was no prior solution to reference. This was a zero-to-one effort.

The Problem

At any point in a deal, F&I managers, dealers, lenders, and customers are each generating and exchanging documents. Without a centralized system, that process was fragmented — documents were scattered, hard to track, and difficult to share. Users had no reliable way to know which documents were required, which were missing, or where things stood at any given moment.

The volume and variety of forms involved — compliance documents, lender contracts, customer-facing agreements — made the problem especially acute. The stakes were high: missing or mishandled documents could delay or derail a deal entirely.

My Process

Stakeholder Interviews. Because this product didn't exist yet, I began by conducting stakeholder interviews to align on scope, priorities, and constraints. Multiple stakeholders — spanning product, compliance, and lender relations — had varying needs and expectations for what the Deal Jacket should do. Getting those perspectives on the table early shaped the design direction significantly.

Personas & Journey Mapping. The Deal Jacket had to serve multiple distinct user types: F&I managers, dealership admins, customers, and lenders. I developed personas for each and mapped their journeys through the deal lifecycle to understand where document needs surfaced, where hand-offs happened, and where friction was highest. This work was foundational — it revealed that the same document could mean something very different depending on who was touching it and when.

Heuristic Audit. Even without an existing Deal Jacket to audit, I conducted a heuristic review of adjacent workflows within Dealertrack to identify established patterns, inconsistencies, and conventions I could build on or deliberately diverge from.

Wireframes, User Flows & Prototyping. I moved from low-fidelity concepts through to high-fidelity prototypes across multiple rounds, designing in parallel for both desktop and mobile experiences. The F&I workflow lives primarily on desktop, but document capture and customer-facing interactions required a thoughtful mobile experience that felt equally native. Prototypes for both surfaces were used directly in usability testing sessions, allowing me to validate navigation patterns, document status visibility, and the core add/edit/share/remove interactions before committing to final designs.

High-Fidelity Design. Final screens covered the full feature set: contextual document lists, automated document generation, forms library integration, manual upload (both desktop and mobile), and secure document exchange with customers and lenders.

Key Design Decisions

Contextual document lists, not static checklists. Rather than presenting users with a fixed list of every possible form, the Deal Jacket dynamically surfaces the documents relevant to each phase of the deal. This reduced cognitive load at a moment when F&I managers are already managing a lot — the right documents appear at the right time, nothing more.

Designing for multiple personas without multiplying complexity. Each user type — admin, F&I manager, customer, lender — interacts with documents differently. The challenge was creating a system flexible enough to serve all of them without making any single experience feel cluttered or over-engineered. Clear permission structures, role-appropriate views, and thoughtful document status communication were central to making this work.

Multiple upload paths, one coherent experience. Documents enter the Deal Jacket in a variety of ways: automatically generated from deal data, pulled from a forms library, or manually uploaded via desktop or mobile. Each entry point had to feel native and consistent — not like four separate features stapled together. Designing for both surfaces also meant rethinking certain interactions entirely; what works as a file picker on desktop becomes a camera-based capture flow on mobile.

Auditability baked in. Given the compliance requirements around deal documentation, every design decision around document exchange and storage was made with auditability in mind. Users needed to trust that the system had a clear, traceable record of what was sent, received, and signed.

Outcome

• Delivered the first-ever centralized document management solution within the Dealertrack F&I platform
• Reduced document-related friction across multiple deal phases for F&I managers, admins, customers, and lenders
• Usability testing validated core workflows across all primary personas prior to development
• Established a scalable design framework accommodating future enhancements including digital signing (OneSpan), ReadySign, and deal document orchestration

Reflection

Designing a document management system from scratch — with no existing product to reference and multiple distinct user types to serve — was one of the more complex scoping challenges of this project. The personas and journey mapping work paid dividends throughout: when design decisions got complicated, I could always come back to a specific user at a specific moment in the deal and ask whether the design was actually serving them. That discipline kept the product focused when stakeholder requests threatened to expand scope in directions that wouldn't have served the core experience. Designing across desktop and mobile simultaneously added another layer of complexity — the two surfaces weren't just different screen sizes, they represented fundamentally different contexts of use that required their own logic.