I build fintech products, and the systems product teams run on.
Head of Product with nine years in fintech and B2B SaaS — most recently at Backflip, a real estate lending platform, where I owned the technology roadmap across three teams and built the AI infrastructure the product org ran on.
My work tends to start with a measurement problem. Teams usually know what they want to build; far fewer can tell you whether the last thing worked. I spend a lot of time on the boring middle — what the metric actually means, whether the pipeline behind it is lying, what the data says about the feature everyone already decided to cut.
Before Backflip I spent six and a half years at Kasasa, a banking SaaS company, moving from Associate PM to Senior PM across rewards platforms, a 600-client platform migration, and a product line that did $1.2M in annual revenue. Before product, I managed a 90-person shift at an Amazon fulfillment center. Purdue, Industrial Management.
Selected work
Making retention the roadmap
Rebuilt the roadmap around new-user retention, then ran a cross-functional pod through 10+ funnel experiments — and showed sales why retention was their problem too.
Saving the product everyone wanted to cut
A feature was on the chopping block for its data bill. The data said it was the single strongest retention driver we had. I kept the product and killed the bill.
Rebuilding a live A/B test that was lying to us
Three variants, two opposed stakeholder agendas, and an attribution pipeline quietly discarding three quarters of real mobile traffic. Fixed mid-flight.
Also
- Loan dashboard architecture. Greenlit a React rewrite and cut the database middle layer in favor of a direct CRM front end — median time to close fell 27% (26 days to 19), engineering feature velocity roughly tripled.
- Data warehouse & securitization reporting. Locked scope at the raw-source and per-loan-derived layers to avoid blocking on downstream logic, inventoried 21 reports across credit, capital markets, accounting and reconciliation, and reduced 500–600 apparent CRM fields to roughly 110 unique source fields.
- Regulatory risk in the funnel. Worked with compliance on denial-of-credit exposure by redefining the application trigger point — an “application” to “loan quote” terminology shift that preserved the funnel and avoided an automated filtering system that would have disqualified good leads.
- Vendor migration off a legacy draw-management platform. Risk-tiered by a 50% loan-completion threshold, with a hard trigger condition that bounded the dual-system period instead of leaving the cutover open-ended.
AI
The last stretch of my work has been building AI systems rather than writing strategy decks about them — a PM operating system on Claude Code, a prototyping pipeline that turns PRDs into working HTML, an event-sourced context architecture for multi-agent writes, and the governance model that let non-engineers ship automations safely.
Contact
Best reached by email.