You have the company. I build the product.

A multi-tenant platform that home care agencies run their day on: shift scheduling, caregiver and client records, applicant tracking, messaging and documents. Four roles — owner, manager, caregiver and family — each seeing a different product.

The part I care most about is invisible: per-tenant isolation enforced at the database layer, so no agency can reach another's records even if the application code is wrong. Security that depends on the UI behaving is not security.

Caregivers clock in and out with GPS verification and the timesheet builds itself, then the platform computes gross, tax and net against current federal and California tables. When two payroll partners declined our applications, I replaced the integration with an in-house calculation engine — which removed the dependency and kept the platform out of the money-transmission path entirely.

A two-sided marketplace connecting homeowners with stone and tile contractors — quoting, bookings, messaging, reviews, and subscription billing driven by Stripe webhooks.

I built a production error-monitoring pipeline that captures browser and backend faults, groups them by normalized signature, and alerts once per issue rather than once per occurrence. Then I did the less obvious half: filtered out the third-party scripts we do not control. Most of the alerts were a bug in someone else's analytics beacon. An alert stream nobody reads is worse than none.

The finding I am proudest of was uncomfortable. Reconciling edge traffic counts against analytics showed our numbers overstated real demand by roughly thirty times — the rest was bots. A strategy had already been built on that number. I would rather kill my own conclusion than defend it.

A consumer savings product built on a serverless edge stack with static generation, holding a 90+ Lighthouse performance score across the site. I own the content architecture, the technical SEO and the automated publishing pipeline behind it.