The next phase of data platforms, picking the best compute engine for any task * talking about pain points with a single warehouse * the entire company needs to work with a single vendor * tied to one engine that might be good for some things but less good for others not the best feedback loop when developing -- always connecting to a server even for small talks * the future of transformation is backed by open data formats and flexible compute * showing how people can build a data mesh with iceberg as the storage format (concepts + demo) * current challenges with iceberg catalogs * how the iceberg ecosystem will likely evolve in the future * improving the developer experience with local compute * showing a demo of a case where a Snowflake dbt project is run locally on DuckDB with automatic conversion of Snowflake dialect to DuckDB -- no compute costs in Dev and a super fast feedback loop
