THE ARCHITECTURE OF DOORDASH'S SEARCH ENGINE

Summary from the TLDR Web Dev newsletter

Elasticsearch was not scaling well for DoorDash’s search engine and lacked customization for the company’s needs in modeling complex data and search ranking, so DoorDash built its own search engine using Apache Lucene, a core search library, and designed its own indexer and searcher to meet its specific requirements. The indexer processes documents into easily searchable formats, while the searcher replicates index data and uses Lucene's search functionality to match queries and rank results.