How do you optimize dbt models for performance when dealing with high-volume, streaming data for Eight Sleep?
