Cafe Niloufer — Local-First AI Inventory Intelligence
Project brief
Autonomous on-premise inventory tracking, waste reduction engine, and predictive demand forecasting.
Overview
Cafe Niloufer is a historic institution serving thousands of cups of legendary Irani chai, Osmania biscuits, and freshly baked confectionery every single day across high-footfall flagship locations.
Managing inventory for high-volume perishable dairy, specialized tea blends, and artisanal baked goods was historically tracked through clipboards and late-night spreadsheets, leading to frequent peak-hour stockouts and preventable spoilage.
Local-first AI architecture
Engineered an air-gapped, local-first inventory management platform running on-premise hardware using quantized open-source LLMs (Llama 3 via Ollama) and Python FastAPI.
Because kitchen back-offices and basement pantries frequently lose reliable cloud internet connectivity, all predictive analytics, batch scanning, and ledger balancing execute 100% locally with offline SQLite sync.
Local-first inventory prediction dashboard and dairy shelf-life tracker
Predictive demand forecasting & smart alerts
Trained custom regression and local AI reasoning agents on 24 months of sales velocity, local weather forecasts, festive calendar peaks, and footfall patterns.
The system generates automated morning prep manifests, alerts head chefs to impending dairy expirations 48 hours in advance, and suggests optimized procurement purchase orders with a single-click export.
Outcome
Kitchen managers cut perishable waste by 31% in the first quarter, while stockouts of signature items dropped to virtually zero during peak Sunday rushes.
Technologies
Built with modern, production-grade tools focused on long-term maintainability, security, and sub-second performance.
Have a project in mind?
Send an e-mail with as much detail as you can about the project and I'll get back to you with availability and an accurate estimate.