SELECTED WORK

Case studies, not project cards.

Context, constraints, decisions and tradeoffs behind the platforms I've built and led - kept qualitative where implementation details are confidential.

Enterprise digital commerce

Platforms operating at national scale

Multi-tenantCheckoutPaymentsLoyaltyCloud-native

Context: A multi-tenant digital commerce platform serving more than 50 U.S. restaurant brands and thousands of physical locations.
Constraint: Each brand needs independent branding and configuration while sharing a common platform core, without duplicating logic per tenant.
Decision: Invest in a shared, configuration-driven multi-tenant architecture rather than per-brand forks.
Architecture: Cloud-native and serverless services handling checkout, payments, authentication and loyalty, layered behind tenant-aware configuration and feature flags, integrated with customer-engagement ecosystems such as Olo, Thanx, Spendgo, Sparkfly, Braze and Klaviyo.
Tradeoffs: Shared architecture raises coordination cost for brand-specific requests, but avoids fragmentation and duplicated maintenance across dozens of codebases.
Result: A platform where operational support work scales sub-linearly with the number of brands onboarded.
What I learned: Multi-tenancy is a discipline enforced at every layer, not a single infrastructure decision.

Accessibility as an engineering practice

Accessibility before QA

WCAGDesign SystemsKeyboardHand TalkEqualWeb

Context: Institutional Bradesco / Bradesco Seguros portals needed to meet accessibility expectations for a broad public audience.
Constraint: Legacy markup patterns and limited automated tooling made systemic accessibility difficult to enforce.
Decision: Treat accessibility as a component-level and workflow-level concern rather than a late-stage QA pass.
Architecture: Semantic HTML foundations, a shared Design System, keyboard navigation patterns, visible focus states, and integrations such as Hand Talk for sign-language support.
Tradeoffs: Retrofitting accessibility into existing components took longer than building it in from the start.
Result: Measurable improvements to keyboard operability, readability and contrast across institutional experiences.
What I learned: Shifting accessibility earlier in the workflow is cheaper than shifting it left after the fact - the premise behind my writing on the topic.

Observability & production reliability

Reliability is a designed property

BugSnagLogglyIncident responseRelease governance

Context: High-volume digital commerce operations require fast detection and resolution of production issues across many tenants.
Constraint: Incidents in a multi-tenant system can silently affect a subset of brands without obvious global signals.
Decision: Build habits and tooling around observability and structured incident investigation rather than reactive firefighting.
Architecture: Error tracking and logging pipelines (BugSnag, Loggly) feeding investigation workflows, paired with release governance and production-readiness review.
Tradeoffs: Observability infrastructure has an upfront cost that competes with feature velocity, but reduces the blast radius and duration of incidents over time.
Result: Faster root-cause identification and more confident, better-governed production deployments.
What I learned: Reliability is a designed property of a system, not a byproduct of careful coding.

Banking & fintech workflows

Correctness over reusability

PIXBoletosBatch paymentsVue / Nuxt

Context: K8 Fintech's banking and financial products needed onboarding, account management and payment journeys built to strict correctness expectations.
Constraint: Financial workflows such as PIX, boletos, transfers and batch payments carry low tolerance for ambiguity or silent failure.
Decision: Prioritize explicit state handling and clear user feedback over generic, reusable-at-all-costs components.
Architecture: Vue/Nuxt front-end journeys integrated with banking and receivables-financing back-office systems.
Tradeoffs: Domain-specific components were less reusable across the product, but produced clearer, safer financial flows.
Result: Reliable onboarding, transfer and batch-payment experiences across banking and HR-adjacent systems.
What I learned: In regulated domains, correctness and clarity outrank component reuse.

Platform modernization

Modernizing a system that can't stop

E-commerceCRMLogisticsPerformance

Context: Atual Card do Brasil's e-commerce, CRM and logistics systems had accumulated years of incremental change.
Constraint: The business needed continuous operation while the platform evolved - a full rewrite was not viable.
Decision: Modernize incrementally, prioritizing performance and the areas with the highest operational impact.
Architecture: Performance optimization, campaign tooling and mobility improvements layered onto existing e-commerce, CRM and logistics systems.
Tradeoffs: Incremental modernization is slower to show dramatic before/after results, but keeps the business running without disruption.
Result: Measurable performance gains and a platform that continued evolving without a costly full rewrite.
What I learned: Sustainable modernization respects the constraint of a system that cannot stop running.

Technical leadership & governance

Repeatable decision-making, not heroics

Tech LeadMentoringRelease governance

Context: As Tech Lead, platform growth required more structured technical decision-making than ad hoc review.
Constraint: Balancing delivery speed with code quality, mentoring needs and production risk across a growing team.
Decision: Establish clear technical steering, code review standards and production-readiness gates instead of case-by-case judgment calls.
Architecture: Architectural reviews, mentoring practices, and a release-governance process that includes production PR and deployment approval authority.
Tradeoffs: Formal governance adds friction to individual autonomy, but reduces variance in production risk across a growing platform.
Result: More predictable releases and a clearer path for other engineers to grow into ownership.
What I learned: Technical leadership is mostly about building repeatable decision-making, not making every decision personally.

Selected brands supported

A representative sample of national and regional restaurant brands operating on the commerce platforms I work on at Plein Air Agency. Listed for transparency about the scale of the work - inclusion here does not imply endorsement by these companies.

goop KitchenLee's Famous RecipeBojanglesLong John Silver'sJason's DeliDig InnCiCisNewk'sTex'sDiBella'sTaco PalenqueCosta VidaCotton PatchNeighborlyTatteO'Charley'sFlame BroilerTijuana FlatsBaja FreshCold Stone CreameryPlanet SmoothieSweet FrogTacoTimeWetzel's PretzelsRusty TacoNekterCaptain D'sFlying PieGolden ChickHello HiloChicken Salad ChickDave's Hot ChickenMellow MushroomTaco Time NWPoke HouseNayaFlanigan'sDirty DoughGenghis GrillBiBiBopBud Long'sCharleysLenny'sWing It On99 RestaurantsAltitudebartacoBoard & BrewFresh BrothersHomeward KitchenJim 'N Nick'sKrispy Krunchy ChickenMain EventRazzoo'sSbarroPuraLimaRapid Fired PizzaSonny's BBQWhataburger