Highlights4
- Adaptive polling: cut driver-location serving cost 45% with poll intervals keyed to distance remaining, interpolated on the client so movement still looks continuous.
- Live order tracking: a Google Maps screen with route polylines and an ETA recomputed on every fix, shipped during a quarter of 25% user growth.
- Notification service: WebSocket and Express.js with backpressure-aware fan-out and duplicate-safe reconnection; after launch, reported CSAT rose 28% and support volume fell 35%.
- Competitor data: a Python scraping toolchain pulling catalogs and reviews into the analytics warehouse on a schedule, normalized to one schema.
More detail
All 5 points
Order tracking
- Interpolated between location samples on the client so a slower poll still renders as continuous movement, which is what made the cost reduction invisible to the user.
- Kept the map honest under a stale fix: the marker holds its last known position with the timestamp shown rather than drifting toward a guess.
Notifications
- Made reconnection safe: exponential backoff plus a per-message id, so a socket that drops mid-delivery resumes without sending the same notification twice.
- Wrote the degraded path deliberately: when the socket is unavailable the client falls back to polling rather than going silent, because a missed order update is a support ticket.
Data
- Automated competitor-catalog harvesting with a Python scraping toolchain, pulling catalogs and reviews into the analytics warehouse on a schedule, normalized to one schema so pricing, positioning and CX workflows read the same shape.
Tools
- React.js
- Google Maps API
- Python
- Node.js
- Express.js