- Predictive dispatch. scikit-learn and K-means demand zones, plus an order-splitting algorithm using traffic, driver load and queue data across pharmacies.
- Odoo Inventory. Custom fields, models, workflows and views, integrated with Sales and Purchase so stock updated itself.
- Data. A PostgreSQL schema of 80+ tables, Elasticsearch catalogue search, Redis caching.
- Rimads. Patient and pharmacist consultation with text, voice notes and medication logs.
- Infrastructure. Owned the backend infrastructure and CI/CD.
PythonDjangoPostgreSQLOdooscikit-learnElasticsearchRedis
Read it left to right: fulfil from one pharmacy when it holds everything, otherwise split across the fewest, then rank drivers by a weighted cost.