dash: delivery planning system

/the idea

built for the tech-triathlon 2026 hackathon, dash is an end-to-end logistics and delivery operations platform developed for waypoint group. the challenge was to connect the full lifecycle—ordering, automated route scheduling, warehouse loading, transit, and retail receipt confirmation—into a single, highly reliable system that works even in complete network dead zones.

/features

• automated constraint-based dispatch: intelligent planning engine that enforces vehicle weight/volume limits, driver working-hour rules, and store delivery windows with instant rule checking for manual drag-and-drop overrides.
• offline-first mobile apps: installable progressive web apps for dock loaders (loading last-stop-first with barcode verification) and drivers (turn-by-turn navigation, location tracking, and offline proof of delivery) that seamlessly operate without connectivity and synchronize when reconnected.
• secure delivery verification: multi-party receipt confirmation using dynamic 6-digit store delivery codes, recipient signatures, and photo capture for discrepancies or shortages.
• waypoint people (hr portal): standalone management panel for staff records, driver license compliance tracking, and granular role-based authentication.

/implementation

the architecture separates operational workloads across distinct services running within a dockerized environment. the backend api proxies requests from both the operations frontend and the hr management panel, backed by postgresql 17 with pglite fallback in development.

for field reliability, the driver and loader interfaces use client-side indexeddb storage and service workers. actions taken offline—such as scanning items, recording delivery arrivals, and collecting signatures—are queued in an idempotent outbox and synced sequentially as soon as network signal is restored, ensuring zero data loss during dead-zone deliveries.

/the tech stack

/thoughts

designing for logistics in regions with patchy cellular coverage highlighted that offline-first is not just a nice-to-have feature—it dictates your entire state model. treating network connectivity as opportunistic allowed us to build software that field operators can genuinely depend on when driving through dead zones.