Dealmate

Dealmate is a real estate lead platform built for real estate owners. I built the full product — the dashboard, the API, the scrapers, the data pipeline, every screen.

The live platform is at dealmate.network.

What it does

Real estate investigators lose weeks finding and vetting seller leads across fragmented public sources — sheriff sale lists, probate filings, tax-lien notices. Dealmate pulls them into one CRM, enriches each lead with owner contact data, and gives pros a single pipeline to work from. For sellers on the other side, it's a no-negotiation quick exit at a pre-agreed price.

The lead pipelines

Three sources feed the database: sheriff sales, probate auctions, tax liens. Each one has its own scraper, its own parser, and its own dedup pass.

Probate is the messiest. Letters of Administration come through as PDFs; case detail pages come through as HTML. Attorney names sit in unstructured text; addresses get cross-referenced against parcel records. A state machine with a 30-day TTL catches duplicates across runs. An atomic lock on the orchestrator stops two scheduled runs from colliding when the cron fires while a previous one is still going.

Once a lead lands, an enrichment job fires in the background. Per-lead enrichment looks up owner names, mailing addresses, and contact details across a couple of providers. A failure from one provider doesn't block the lead — partial data shows as a badge instead of an error, and a retry button stays available. The frontend subscribes to row changes, so progress lights up in the UI without polling.

CRM surface

Kanban board for the deal pipeline with drag-drop status changes. Drawer-based create and edit so the list view stays put while you're working a lead. Filters across status, source, and recency that collapse out of the way when you don't need them. CSV export with phone-number dedup and clean handling of nulls. Pagination at twenty-five leads per page so the table never gets out of hand.

Auth and admin

Two surfaces — admin dashboard and dealer dashboard — each with its own auth flow and route guards. Service-role access on the backend gates privileged operations behind role checks. The scraper service runs scheduled pipelines daily and exposes admin-only endpoints with separate token validation. Dev-mode auto-detection skips startup pipeline runs on hot reload so you don't burn through scrape quotas while iterating.

Scope

Roughly 320 commits across four months. Counting the dashboards, the scraper service, the parsers, generated types, and tests, the codebase clears 650,000 lines. More than 80 user-facing routes split between the dealer and admin sides. 25 API endpoints. 16 backend test suites covering parsing, scraping, enrichment, auth, and orchestration. Sixty-eight database migrations as the schema grew from MVP to multi-source platform. Every module — frontend, backend, scrapers, deployment — mine.