AI-ML Pharmacy Stock Management System -- 2

I want to build an end-to-end medicine management platform that lets our pharmacy team stay on top of inventory levels and expiry dates while also giving patients a convenient companion app for their day-to-day medicine routines. Core workflow • Inventory & expiry control – intelligent forecasting, batch tracking, low-stock alerts and automated removal of soon-to-expire units, all surfaced in an easy dashboard for pharmacy staff. • Patient-facing tools – the same database powers three functions: dose reminders with in-app confirmations, digital prescription uploads with automatic refill prompts, and real-time interaction warnings that feed into personalised health reports. • AI/ML layer – models (TensorFlow / PyTorch are fine) analyse historical sales and seasonality to predict demand, flag anomalies and continually optimise reorder points. Tech preference A modern stack such as Python + Django (or Node) for the API, PostgreSQL for secure data, and React Native or Flutter for the patient app will keep it maintainable. If you have a better architecture in mind, I’m open, as long as the AI models remain modular and the system can scale across multiple pharmacy branches. Deliverables 1. Source-controlled codebase with README and deployment scripts. 2. Trained ML models with documented datasets and retraining pipeline. 3. Web dashboard for pharmacy staff and a cross-platform mobile app for patients. 4. API documentation (OpenAPI / Swagger). 5. Test suite and final hand-over demo on our staging server. Acceptance criteria • Stock-out prediction accuracy ≥ 90 % on three months of back-tested data. • Expiry alerts triggered at configurable lead times and reflected in UI within 60 s. • Push notifications delivered reliably on both iOS and Android (>95 % success). • All features usable behind role-based access and passing OWASP Top 10 checks. This is a green-field build, so you will have freedom to choose libraries and cloud services as long as they meet the above criteria and keep the code portable. Let me know how you would tackle the data pipeline first, and we can break the work into milestones from there.