À la carte — Modern Kitchen Operating System
Overview & Objective
À la carte unifies everyday household food logistics into a single real-time collaborative platform. Designed from the ground up around a private Family circle where all household members share synchronized access to pantry stock, menus, and shopping carts. Features a multi-modal recipe ingestion pipeline (manual entry, Gemini AI Chef suggestions from fridge ingredients, URL web scraping, and computer vision camera OCR scanning physical cookbook cards), preloaded with 108 verified recipes across thematic Starter Packs. Engineered as an offline-capable PWA with React 18, Supabase PostgreSQL Row-Level Security (RLS), and Deno serverless edge functions.
System Architecture & Modules
┌────────────────────────────────────────────────────────────────────────┐
│ Frontend Client (React 18 + Vite) │
│ Tailwind CSS • Radix UI (shadcn) • TanStack Query • Zod Forms │
└───────────────▲───────────────────────────────────────▲────────────────┘
│ │
│ Supabase JS Client │ HTTP Edge Calls
│ (PostgreSQL + Auth + Storage) │ (Bearer JWT Auth)
▼ ▼
┌──────────────────────────────────────┐ ┌──────────────────────────────┐
│ Supabase Backend Service │ │ Deno Edge Functions │
│ │ │ │
│ • PostgreSQL with RLS Policies │ │ • suggest-recipe (Gemini AI) │
│ • public.is_shared_with_me() │ │ • scan-recipe-image (Vision) │
│ • GoTrue Authentication │ │ • scrape-recipe (Web Parser) │
│ • Supabase Storage (Recipe Photos) │ │ • invite-family (Mailer) │
└──────────────────────────────────────┘ │ • stripe-webhook (Stripe) │
└──────────────▲───────────────┘
│ External APIs
┌──────────────▼───────────────┐
│ Google Gemini • Stripe API │
└──────────────────────────────┘Multi-Modal Recipe Ingestion Engine
Adding recipes to family menus is frictionless through four parallel creation pathways:
AI Chef (Google Gemini API)
Analyzes current ingredients in the pantry and generates balanced, custom recipes on demand via the suggest-recipe Deno Edge Function.
OCR Vision Scanner
Uses computer vision to photograph physical cookbook pages or handwritten recipe cards and convert them into structured ingredients and steps.
One-Click URL Web Scraper
Parses culinary blogs and cooking sites into normalized schema data, discarding ads, trackers, and fluff.
108 Preloaded Verified Recipes
Includes structured Starter Packs for Quick & Easy, Student/Budget, Vegetarian, Family Meals, Batch Cooking, and Pastry.
Database Architecture & Row-Level Security
Data security and multi-tenant separation are enforced strictly at the database engine level via PostgreSQL Row-Level Security (RLS). All rows in recipes, stock, and shopping_list are verified using the security function public.is_shared_with_me(item_owner_id uuid), ensuring complete tenant isolation without application-layer leaks.
Key Engineering Highlights
Real-Time Collaborative Family Hub
Private family workspace with instantaneous cross-device synchronization powered by Supabase PostgreSQL Row-Level Security (is_shared_with_me).
Multi-Modal Recipe Ingestion Pipeline
Import recipes in seconds via Google Gemini AI suggestions from available stock, camera OCR scan of physical recipe cards, or one-click URL web scraping.
Aisle-Optimized Smart Shopping Lists
Interactive grocery lists automatically organized by supermarket aisles with optimistic local cache updates via TanStack Query.
Kitchen-First Progressive Web App (PWA)
Installable on iOS, Android, and desktop with offline support, bottom drawers (Vaul), and touch-first kitchen ergonomics.