VerdeAI

🌱 VerdeAI AI‑powered food‑waste reduction platform for restaurants

VerdeAI is a serverless, cloud‑native system that helps restaurants reduce food waste through demand forecasting, waste‑risk scoring, and natural‑language recommendations. The platform combines Azure Functions, Cosmos DB, and Azure OpenAI to turn raw operational data into actionable insights.

🚀 Features Demand Forecasting
Predicts daily item‑level demand using historical sales and contextual signals.

Waste‑Risk Engine
Scores each menu item for overproduction risk and identifies potential waste hotspots.

AI‑Generated Explanations
Uses Azure OpenAI to convert forecasts and risk scores into clear, manager‑friendly insights.

Serverless Architecture
Fully event‑driven pipeline built on Azure Functions for scalability and low operational overhead.

🏗️ Architecture Overview Code +------------------+ +------------------+ +------------------------+ | Restaurant | | Azure Cosmos | | Azure OpenAI | | Data Source +------->+ DB +------->+ (verdeai-explainer) | +------------------+ +------------------+ +-----------+------------+ | ^ | v | v +------------------+ +------------------+ +------------------------+ | VerdeAI_Forecast | | VerdeAI_Waste | | VerdeAI_Explain | | (Azure Function) | | (Azure Function) | | (Azure Function) | +------------------+ +------------------+ +------------------------+ 📦 Tech Stack Azure Functions (Node.js)

Azure Cosmos DB

Azure OpenAI Service

Event‑driven serverless pipeline

HTTP‑triggered APIs for dashboard integration

🛠️ Local Development

1. Install dependencies Code npm install
2. Configure environment variables Create a local.settings.json file in the project root:

json { "IsEncrypted": false, "Values": { "AzureWebJobsStorage": "UseDevelopmentStorage=true", "FUNCTIONS_WORKER_RUNTIME": "node", "COSMOS_DB_CONNECTION_STRING": "", "OPENAI_ENDPOINT": "", "OPENAI_API_KEY": "", "OPENAI_DEPLOYMENT": "verdeai-explainer" } } 3. Start the Functions runtime Code func start 📘 Function Endpoints

1. VerdeAI_Forecast Code POST /api/VerdeAI_Forecast Generates item‑level demand forecasts.

2. VerdeAI_Waste Code POST /api/VerdeAI_Waste Computes waste‑risk scores based on forecasted demand.

3. VerdeAI_Explain Code POST /api/VerdeAI_Explain Returns natural‑language insights for managers.

Example request:

json { "restaurantId": "R001", "date": "2026-02-05" } 📊 Output Example json { "item": "Grilled Chicken Bowl", "forecast": 42, "wasteRisk": "High", "explanation": "Demand is expected to drop today due to lower weekday traffic. Reduce prep volume by ~15%." } 🧪 Testing Use curl or Postman:

Code curl -X POST http://localhost:7071/api/VerdeAI_Explain
-H "Content-Type: application/json"
-d '{"restaurantId":"R001","date":"2026-02-05"}' 📈 Impact Improved forecast accuracy by ~25%

Reduced pipeline latency by 40%

Helped pilot restaurants cut food waste by 15–20%