The engineering record.

I joined early enough to sit on the committee that designed the enterprise data architecture — data lake, EDW, reporting layer — as its reporting architect, and helped build the core reporting solution the company scaled on. Later I inherited a half-built CDC/replication platform when its senior engineer left; I shipped it, re-architected it onto Kafka on Kubernetes ($30K+/yr saved), and it's been mine end to end since.

The platform's most consequential layer now is semantic: I created the Claude plugin marketplace — architected it, authored the domain skills, and run its roadmap, evaluations, governance, and support. Business leaders at every level, C-suite included, build QBRs, reserve reviews, and financial reports from it. Everything ships as code, observability included.

The leadership half of the job: code reviews, mentoring, office hours driving adoption of AI-assisted development — serving every level of the business and translating in both directions. Along the way I also built a key component of the enterprise pricing tool (a TypeScript microservice on Azure — GraphQL, MCP, PostgreSQL), wrote SOC 2 SOPs and supported audits, and put adoption decisions on usage analytics.

The clearest reading of the quality bar is public. Waybill is an MIT-licensed Claude Code plugin for token accounting on AI-assisted work — deterministic attribution, conservation checks, verification packs a recipient can re-run offline — with the code and docs in the open, so the standard is inspectable, not asserted. There's also a side project on the App Store: QuotedAI, a SwiftUI/HealthKit app with on-device AI and 2K+ downloads. The phone is a strict teacher of constraints — memory, battery, offline — and every lesson transfers straight back to platform work.