Discover Restaurants Intelligently
CuisineAML combines menu digitization and predictive analytics to help employees in Madrid's Azca zone make faster, smarter lunch decisions.
The Challenge
In Madrid's Azca zone, a bottleneck effect impacts daily dining choices
27,000 Employees
All eating lunch simultaneously between 14:00-16:00 in the same zone
~70 Restaurants
Limited dining options creating unavoidable waiting queues
Menu Repetition
Employees forced to eat at the same restaurants, losing satisfaction
Manual Menus
Restaurants print daily menus on paper with no digital access for customers
The Bottleneck Effect
With 27,000 employees and only ~70 restaurants, there's a severe bottleneck. Employees can't browse restaurants freely—they rely on habitual choices due to time constraints and queue anxieties.
This created a need for a smart discovery and demand prediction solution.
Our Solution
A two-phase AI platform for intelligent restaurant discovery and demand forecasting
Restaurant Intelligence Platform
Restaurants upload daily menu photos. Our system automatically extracts, classifies, and stores menu items. Users can:
- ✓Filter restaurants by location, cuisine, or specific dishes
- ✓Rate dishes and see community rankings
- ✓Compare today's menus in seconds
- ✓Discover new favorite restaurants
AI Demand Forecasting
Using historical data and advanced ML models, we predict:
- 01Daily service volume per restaurant
- 02Weekly menu recommendations
- 03Peak dining times and congestion
Restaurants receive actionable insights to optimize staffing, inventory, and menu planning.
Product Demo
Short walkthrough of CuisineAML in action, covering discovery, catalog filtering, and restaurant-side workflows.
Key Features
Comprehensive platform for discovery and prediction
Menu Photo Upload
Restaurants upload daily menus via photo—no complex systems needed
AI Classification
Document Intelligence automatically extracts and categorizes dishes
Smart Search & Filter
Filter by cuisine, price, location, dietary preferences, and ratings
Community Ratings
Users rate dishes to build a trustworthy ranking system
Demand Prediction
XGBoost ML model predicts daily service volumes with 30+ features
Weather Integration
Real-time weather data improves prediction accuracy
Calendar Features
Holiday and payroll periods considered in forecasting
Location Intelligence
Geolocation-based restaurant recommendations
System Architecture
Production-ready infrastructure built on Azure
Frontend
- ✓ React + Vite
- ✓ Tailwind CSS
- ✓ Mobile-responsive
- ✓ Real-time updates
Backend
- ✓ FastAPI + Python
- ✓ SQLAlchemy ORM
- ✓ RESTful APIs
- ✓ Audit logging
ML Pipeline
- ✓ XGBoost model
- ✓ AutoML Studio
- ✓ 30+ features engineered
- ✓ Azure Logic Apps
Data Sources & Enrichment
Azure SQL Database
Restaurant details & historical services
Open-Meteo API
Real-time weather data
Calendar Features
Holidays & payroll periods
Historical Data
Service volumes & trends
System Flowchart
End-to-end flow of data ingestion, enrichment, prediction, and user-facing output.
Technology Stack
Modern technologies and Azure cloud services
Frontend
- React 18.3
- Vite 5.6
- Tailwind CSS 3.4
- TypeScript
Backend
- FastAPI 0.135
- Pydantic 2.12
- SQLAlchemy 2.0
- Python 3.10
Database
- Azure SQL Server
- pyodbc
- SQLAlchemy ORM
ML & AI
- XGBoost 1.5
- Azure AutoML
- Feature Engineering
- Azure Machine Learning
Data & APIs
- Open-Meteo API
- holidays library
- Document Intelligence
Infrastructure
- Azure SQL
- Logic Apps
- Uvicorn Server
- Azure Functions
30+ Engineered Features for ML
4 Data Sources Integrated
100% Production Ready
Explore Screenshots
All project visuals with descriptive navigation and enlarged preview
Landing Discovery
1 / 11Main discovery view with search, category pills, and premium positioning.
Meet the Team
Individuals behind this project
Contributors
Completion Year