I spent ten years underwriting credit and lending before teaching myself to build software. I then designed and delivered the production lending platform my company operates on — independently, from architecture to deployment. I bring both the domain expertise and the technical capability to roles where finance and technology meet.
I am a credit underwriter by training, with a decade spent analyzing bank statements, cash flow, and fraud indicators to determine which businesses receive funding and on what terms. This represents deep domain expertise in a vertical few engineers understand firsthand.
When my company required software to operate on, I built it rather than outsourcing it. I taught myself Apex, Lightning Web Components, SOQL, Supabase, and the Claude API, and delivered the complete system: a production CRM and lending platform managing real deals, capital, and underwriting decisions.
This combination — domain expert and capable builder — is what I bring to roles where finance, technology, and the customer intersect: understanding the problem in the customer's terms, then helping deliver what solves it. I have been both the end user and the engineer.
A production Salesforce-based lending platform that operates a full merchant cash advance business end to end — live software handling real deals, payments, and underwriting decisions. The following are four of the core systems within it.
A configurable rules engine that scores each deal across more than twelve risk signals — time in business, FICO, revenue, deposits, negative days, existing positions, and bank type — then assigns a risk tier or flags it for manual review with a complete audit trail. It triggers automatically when a deal enters underwriting.
Bank-statement PDFs are submitted; structured underwriting data is returned. The Claude API extracts deposits, balances, NSFs, and negative days, automating a process that previously required manual line-by-line review. Built asynchronously to respect platform processing limits.
Role-gated portals for syndicators, investors, and accountants, each with access limited to their own data. Monthly payout settlement, ROI and performance views, and server-generated PDF and CSV reporting, all behind authenticated access.
Data models, triggers, queueables, processing-limit management, and a decline workflow with 27 categorized reasons and automated broker notifications. Related-submission detection matches repeat applicants by normalized SSN. 248 passing tests in production.
A brief screen-recorded tour of the platform — the classifier, the document extraction, and the portals in operation.
Loom walkthrough — coming soon
If you are building at the intersection of finance and technology and seeking someone who understands both the domain and the customer, I would welcome the conversation.