Case Study: Hopper visualizes 4.4 billion flight prices with Union.ai

Hopper visualizes 4.4 billion flight prices in one day with Union.ai

Hopper, a global travel platform, faced the significant challenge of visualizing 4.4 billion flight prices in an interactive map without getting bogged down by the painful debugging and deadlocking issues common with distributed processing frameworks like Ray. They needed an orchestration solution that provided algorithmic control without the associated infrastructure complexity to build their sophisticated Deep Zoom visualization.

Using Union.ai's pure Python orchestration, Hopper implemented a MapReduce-style shuffle algorithm with async-native patterns, achieving a production-ready solution in just one day of development time. The Union platform handled auto-scaling across 7,000 concurrent executions to process the 4.4 billion trip records efficiently. This resulted in a scalable workflow that worked on the first attempt with the full dataset and provided an end-to-end pipeline from data processing to deploying a served FastAPI application for the visualization.

Hopper