Data Forge owns the data lake and builds the reusable, production-grade datasets that turn raw company signals into reliable data products for ML, search, analytics, and product teams.
We build on shared infrastructure owned by our sister teams. What we own is the lake itself and what lives in it: the storage architecture, catalog, schemas, SLAs, and the lifecycle of every curated dataset we deliver. Our work shows up in conversational search, home recommendations, agent matching, clickstream analytics, and the ML ranking and relevance features behind Redfin's core product.
Three strategic bets define our roadmap. First, Fill the Lake: we are systematically bringing every meaningful raw signal at Redfin into a single, well-governed data store - migrating legacy systems, onboarding new source systems, and expanding coverage so every team builds on the same underlying data rather than maintaining their own copies. Second, unification of our Snowflake and data lake environments: we are converging on a single Iceberg lakehouse where the same tables are queryable through SQL, Spark, and ML tooling - no split schemas, no duplicate copies for different consumers.
Third, accessibility: we are reducing the friction between raw lake data and the feature teams who need it, building the data contracts, discovery patterns, and self-service interfaces that let ML engineers and product teams find, trust, and use lake data without needing a Data Forge engineer in the loop.
We are building op