PAZZ Marketplace · Technical Lead · Current work
Rebuilding automotive leasing as one operable marketplace.
Automotive leasing is not one form or one approval. My role is to make its customer, partner, provider, and operator handoffs behave as one coherent product—without hiding the business decisions or human judgment that still matter.
One customer goal, many operating systems.
The customer sees a simple goal: find a vehicle and understand whether leasing it makes sense. Behind that goal are vehicle data, pricing and tax rules, eligibility, documents, partners, external systems, and human review.
Improving one screen in isolation would not solve the system. Catalog data must be usable. Quotes must apply the same rules everywhere. Prequalification must preserve context. Operators need to know what happened, what failed, and what to do next.
Technical direction that remains hands-on.
I set technical direction and remain active in implementation across product, engineering, operations, and external constraints.
- Product architecture and delivery sequencing
- Customer-facing web and mobile experiences
- Vehicle discovery, quoting, and prequalification
- Conversational intake and bounded applied-AI workflows
- Partner, operator, document, and contracting handoffs
- Shared domain models, data, integrations, observability, and release gates
Selected engineering decisions.
The work treats the marketplace as one operating path rather than a collection of disconnected interfaces.
- Centralize quote and eligibility semantics instead of duplicating them across customer, partner, and operator surfaces.
- Preserve context through multi-step journeys so re-entry does not restart the process or discard prior answers.
- Model handoffs as owned states with explicit inputs, outputs, and recovery behavior.
- Keep models inside bounded workflows rather than allowing them to become silent business authority.
- Treat timeouts and partial failures as expected operating states, not exceptional afterthoughts.
- Separate source, CI, runtime, and production evidence, then bind release claims to the artifact actually verified.
Evidence boundary.
PAZZ operates publicly at pazz.mx, while the production platform and most of the work described here are private.
This case study intentionally excludes customer data, supplier identities, private architecture, internal incidents, work still in flight, security details, and business metrics. It describes verified scope and selected decisions. It does not claim that every workflow is fully automated or qualified through the entire journey, and it does not attribute revenue, conversion, approval, or customer-satisfaction outcomes without an approved measurement record.
Representative stack.
TypeScript · Next.js · React Native / Expo · PostgreSQL / Supabase · GitHub Actions · Sentry · applied AI