Inside Zepto’s 10-Minute Delivery Engine Kafka, ClickHouse, and the Real-Time Tech | Point Break
AI Summary
In this episode of Point Break, we explore the intricate data infrastructure behind Zeppto’s promise of delivering groceries within 10 minutes. Interviewing Abhinav Raguani, the associate director of data at Zepto, we delve into the transition from an outdated data stack on Amazon Redshift to a modern lakehouse architecture built on Delta Lake and powered by Databricks. Zeppto captures and processes vast amounts of data in real-time to optimize delivery speeds while balancing costs and system performance. The discussion highlights how Zeppto distinguishes between data latency types, implements a no-code ETL framework, enforces budgetary controls, and uses ClickHouse for ultra-fast querying to handle critical metrics. The convergence of speed, precision, and data governance is emphasized as essential for maintaining customer satisfaction in a highly competitive market.