Work

Led a 25-person cross-functional team through a full-scale platform migration where the primary constraint was zero degradation during peak traffic. Black Friday/Cyber Monday volumes would expose any reliability gap immediately.

Designed the migration sequence to isolate failure modes: CMS, payment infrastructure, and design system rebuilt in stages, each evaluated for stability before the next phase began. Platform reliability improved 30%, measured against incident frequency and resolution time.

Managed $1.5M in vendor contracts with structured evaluation criteria for selection and renewal. Technical documentation from discovery through delivery reduced engineering rework by establishing clear integration boundaries.

Led product for Arc XP's expansion into new verticals. The evaluation problem was fitness for context: each vertical had different content models, user behaviors, and integration requirements. Vertical-specific demo environments let prospects evaluate the platform against their own constraints before purchase, reducing qualification mismatch and improving lead quality 17%.

Deployed AI/ML recommendation models across Arc XP's global publisher network. Aggregate model accuracy was not the evaluation criterion that mattered. The relevant question was whether recommendations performed reliably for publishers with specific audience distributions and editorial standards. A model that performs well on average but fails for a given publisher's content structure erodes client trust faster than no recommendation system.

Launched the powersports industry's first end-to-end eCommerce checkout platform: digital purchase agreements, trade-in valuations, financing, payments, and 50-state DMV titling. The product was defined more by its constraints than its features: state-by-state regulatory compliance, lender integration requirements, dealer workflow compatibility, and consumer protection obligations.

Constraint mapping was the core product discipline. Each state's titling requirements imposed different validation rules, document formats, and processing timelines. The system had to handle these variations without exposing complexity to the end user or creating failure modes at state boundaries.

Athletic Greens is a subscription business where the most consequential failure modes don't announce themselves. A billing error or a friction point introduced during a platform migration doesn't generate an immediate incident. It generates a customer who doesn't renew.

Evaluated and selected subscription platform enhancements against criteria centered on payment reliability, failure handling at scale, and integration stability. Structured vendor selection around failure modes rather than feature sets. Led CMS implementation under a continuity constraint: the subscription funnel was active throughout, and new content capabilities had to integrate without introducing failure points at the boundary between content and billing systems.

Built and institutionalized a company-wide experimentation program, A/B testing and multivariate evaluation across the full purchase funnel. The 10% conversion lift was the outcome. The system's real value was structural: product decisions were gated by evidence rather than opinion, and every change was evaluated against specific performance criteria before full deployment.

Designed product governance frameworks that defined how decisions were made, who had authority to approve changes, and what evidence was required to ship.

Developed market-level business intelligence tools using ML models to recommend optimal rental rates by ZIP code. The evaluation challenge was calibration: a pricing recommendation is only useful if property managers trust its outputs, which required demonstrating accuracy against local market conditions, not aggregate model performance across all markets. A tool that is right on average but wrong for a specific property manager's market erodes trust in the platform.

Launched customer reporting and analytics infrastructure. The design constraint was interpretability: analytics outputs had to be legible to property managers without data science backgrounds. The system had to present conclusions with enough context to be evaluated by the user, not just consumed.

Led a company-wide conversion rate optimization strategy that functioned as a continuous evaluation architecture. No product change reached full deployment without structured A/B testing against defined success criteria. The conversion lift and performance improvement were outputs of the system. The system itself was the discipline of evaluating before shipping.

Managed a company-wide rebrand and eCommerce platform overhaul. The defining constraint was continuity: content publishing workflows, SEO equity, and active sales flows could not degrade during the transition. Every migration phase was evaluated against regression criteria. The new system had to match or exceed the old system's performance before the next phase began.

Grew organic traffic 25% and online sales 15% through structured SEO/SEM programs and continuous CRO testing. Reduced launch timelines 20% by identifying failure points in the launch sequence earlier, before they became customer-facing problems.