Transportation Roster Platform

Pain Points

• Shift conflicts discovered only when they caused operational disruptions
• Last-minute changes communicated by phone, leading to missed updates
• No visibility into workforce availability or location
• Unfair shift allocation creating staff dissatisfaction
• Reporting on workforce utilization was manual and weeks delayed

Project Goals

• Centralize roster planning in a single platform
• Provide real-time visibility to all stakeholders
• Automate shift conflict detection
• Support fair shift allocation rules
• Enable last-minute changes with instant notification

Requirements Gathering

Worked with transportation department managers to map scheduling workflows. Discovered that shift rules were complex and varied by role and location. This drove the decision to build a configurable rules engine rather than hard-coded scheduling logic.

Architecture Decisions

• NestJS with separate modules for scheduling, roster, and workforce
• React with drag-and-drop roster management for intuitive scheduling
• PostgreSQL with Prisma for typed roster data access
• WebSocket for real-time schedule updates
• Configurable rules engine for shift constraints

Database Schema

API Design

• RESTful roster management endpoints
• WebSocket channel for real-time schedule updates
• Rules engine API for shift conflict validation
• Notification API for schedule change alerts

Authentication Strategy

JWT authentication with RBAC across scheduler, manager, and staff roles. Staff roles have read-only access to their own roster. Schedulers have full roster management access.

Technology Selection

• NestJS — modular architecture for scheduling domain complexity
• React — drag-and-drop libraries for intuitive roster management
• PostgreSQL — relational data model fits roster scheduling well
• Prisma — type-safe data access reducing scheduling query errors
• WebSocket — essential for real-time operational visibility

Trade-offs

• Configurable rules engine over hard-coded rules: more complex but handles diverse scheduling needs
• WebSocket over polling: better real-time experience but more infrastructure
• Drag-and-drop over form-based scheduling: better UX but more frontend complexity

Performance Optimizations

• Optimized roster query performance with database indexing on staff and date
• Real-time updates via WebSocket batched to avoid flooding
• Cached workforce availability data with event-driven invalidation
• Background workers for roster report generation

Security Considerations

• JWT authentication with role-based access control
• Staff can only view their own roster — no cross-staff visibility
• Audit logging for all schedule changes
• Input validation on all scheduling endpoints

Scalability Planning

WebSocket connections can be scaled with a pub/sub adapter. Roster queries can use read replicas. Background workers scale horizontally for report generation.

Deployment Strategy

Dockerized application deployed via CI/CD. WebSocket connections require sticky sessions or a pub/sub adapter. Blue-green deployment for zero-downtime updates.

Lessons Learned

• Scheduling rules are more complex than they appear — model them explicitly
• Real-time updates are essential for operational software
• Fairness in shift allocation is a product feature, not a nice-to-have
• Drag-and-drop scheduling dramatically reduces scheduler training time

Future Improvements

• AI-assisted optimal scheduling with constraint satisfaction
• Mobile app for staff to view and request shift changes
• Integration with vehicle tracking and maintenance systems
• Predictive workforce demand forecasting