A full-stack social marketplace for animal breeders. Built in-house at SDS. Live in pre-beta.
Scrlpets is what you'd get if Instagram, Shopify, and a professional breeder management tool were built as a single product instead of duct-taped together from five different subscriptions.
It's a social feed where breeders post their animals, a marketplace where verified buyers can inquire and purchase, a DM system with real-time chat, a trust layer that surfaces credibility signals, a dashboard where breeders run the operational side of their business — lineage, health records, contracts, payouts — and a set of AI tools that do the work that used to eat their evenings.
The target user is a hobbyist or professional breeder managing 1–50 animals, across nine species — dogs, cats, rabbits, reptiles, birds, insects, fish, exotics, and the long tail. The industry today runs on Facebook groups, spreadsheets, Venmo, email, and whatever app someone's cousin recommended. Scrlpets is one login.
It's live at scrlpets.lovable.app.
These are the numbers because the question a consultancy's portfolio page should answer isn't “is this pretty?” — it's “can these people actually ship something at production scale?” This is our answer.
Frontend
React 18 · TypeScript · Vite 5 · Tailwind CSS · shadcn/ui · TanStack Query · TanStack Virtual
Backend
Supabase · Postgres · Row-Level Security · Edge functions in Deno/TypeScript
Payments
Stripe Connect — escrow, subscriptions, marketplace payouts
AI
Lovable AI Gateway (Gemini 2.5 Flash) across 21 tools · OpenAI Whisper for voice-to-listing
Maps
Mapbox GL JS v3
Delivery
PWA with service worker, offline support, and push notifications
No client-side database. No ORM the database didn't ask for. No “we'll figure out auth later.” RLS is on from day one.
01
A TikTok-style feed that scrolls through thousands of listings can't be built the naive way and patched later. TanStack Virtual was in the architecture from commit one, so the feed only renders what's on screen regardless of how many rows the database has. The perf budget is stable as the product grows.
02
Most Vite projects ship one big vendor chunk and move on. Scrlpets splits vendor code into eight deliberate chunks, because an earlier auto-chunking pass produced a subtle React dispatcher mismatch when two chunks loaded out of order. Fixing it required understanding why the default bundler behavior can break modern React — not just turning knobs until it worked.
03
Rolling back an AI-assisted build where a single bad session can touch thirty files isn't something you handle with git reset. Every stable state gets tagged and documented. The cost is two minutes per checkpoint. The savings are measured in days.
04
131 tables, all governed by Row-Level Security policies from the start. No 'we'll lock it down before launch.' Security that's retrofitted at the end of a build is security that doesn't exist; security that's designed in from the schema up is the only kind that holds.
Scrlpets is in pre-beta. The platform is live, the build is passing, and the current work is onboarding email infrastructure, analytics instrumentation, and beta cohort recruitment for the exotic-breeder segment.
There are no paid users yet. There is no MRR to report. When those numbers exist, this page will be updated with them — not before.
We'd rather show you what we built than make up what it did.
Tell us what you're trying to ship. We'll tell you what it takes.