Software Development Project Demonstration

Context-Aware Next Best Action Platform A full-stack pharmaceutical field-force platform that combines territory management, doctor and pharmacy intelligence, offline-capable visit execution, pharmacy sales signals, and explainable next-best-action recommendations. The system is designed for a realistic medical representative workflow: Managers define territories, assign MRs, assign doctors, import pharmacies, and monitor coverage. Sales reps capture pharmacy orders and product movement. MRs review ranked next-best-action recommendations, plan visits, capture GPS and feedback, and sync offline work. Managers and admins review performance, audit logs, recommendation explanations, and governance metrics. Tech Stack Frontend: React 18, Vite, TypeScript, React Router, PWA service worker. Backend API: Java 21, Spring Boot, Spring Security OAuth2 Resource Server, Liquibase. Recommendation service: Python FastAPI. Data: PostgreSQL with PostGIS. Identity: Keycloak realm import with demo users and realm roles. Deployment: Docker Compose for local orchestration. Prerequisites Docker Desktop Node.js 18+ and npm Java 21 and Maven, if building the API outside Docker PowerShell on Windows, or equivalent shell on Linux/macOS The demo database includes: Core roles: MR, SALES_REP, MANAGER, ADMIN. Demo users and Keycloak realm mappings. Colombo Central territory. Walasmulla Area territory assigned to MR One. Demo doctors, pharmacies, products, visits, pharmacy orders, pharmacy feedback, and recommendation feedback. Product assignments and sales targets for manager/sales workflows. Additional territories can be created from the UI. Implemented software support: Doctor, visit, territory, schedule, GPS, pharmacy, sales, feedback, and product-assignment domains. PostGIS-backed proximity queries for doctors and pharmacies. Role-based access control for MR, sales rep, manager, and admin workflows. Explainable recommendation factors and recommendation feedback. Offline queueing and sync conflict handling. Manager/admin evaluation and governance endpoints. Research work still required outside the codebase: