FireLake is a hybrid search system that combines the strengths of data lakehouse architectures with vector databases to support both structured (SQL-style) and unstructured (semantic) queries. It employs a medallion data pipeline, progressing from raw data inges- tion (Bronze) to refined data (Silver) via automated entity extraction,and finally to enriched analytics tables (Gold). The system’s query engine unifies SQL filtering with vector similarity search, allowing users to query numerical data and text content in a single inter- face. full paper: https://drive.google.com/file/d/1iPfBie6V8wN0YPMmjR2g_w57zm_...