Lovable Web App Code Review and Backend Optimisation

The platform has been built using Lovable. The product is good, but as is common with AI-assisted builds, there are inconsistencies, fragile flows, and a few areas that need a competent human developer to properly diagnose and stabilise before we scale up marketing and onboarding. This is currently a one-off paid review and fix engagement, with the strong possibility of ongoing work if it goes well. I am looking for someone reliable who can work independently, communicate clearly in plain English (I am not a developer), and leave the codebase better documented and more stable than they found it. The platform needs to work at scale and run as automated as possible. Tech stack Frontend: React, TypeScript, TanStack Start/Router, Tailwind CSS Backend: Supabase (Postgres, Auth, Storage, Edge/server functions), hosted via Lovable Cloud Payments: Stripe Maps/Geocoding: Mapbox GL JS + Mapbox Geocoding API Email: Resend + Lovable Cloud Custom Emails (for auth emails) Error monitoring: Sentry (already connected) Hosting: Lovable Cloud (Cloudflare Workers under the hood), custom domain via Cloudflare Source control: GitHub (connected, full repo access available) Business sign-up flow The 3-step registration wizard has had recurring issues with field validation, address/location confirmation, and the post-registration redirect to the dashboard breaking repeatedly. Business dashboard Profile editing, hours, amenities, photo uploads, classes/events, jobs, messaging, and the subscription/billing tab all need a functional review to identify what's reliable versus flaky. Admin panel The internal dashboard for managing users, businesses, leads, and claims has had rate-limiting and data-loading issues and needs a security and performance review at the current scale. General codebase health A broader pass on Supabase RLS security policies, silent failures (things that look like they work but don't, like emails or analytics), and an honest opinion on whether the architecture holds up as the platform scales.