Why it matters
Snowflake Cortex and Databricks expose managed LLM inference endpoints. Teams deploying RAG on them face real decisions about model routing, embedding costs, and retrieval-pipeline token budgets.
The tokenmaxxing angle
Cortex Search uses hybrid vector + keyword retrieval. The talk covers how metadata modeling shapes the semantic layer — directly relevant to minimizing re-ranking passes and controlling embedding token spend at scale.
From the organizers
Speaker Tully Taylor is Manager of Data Science at Lumen Technologies; Michelle Gaedke is Solutions Engineering Manager at Datalere. Denver MLOps community, evening format with networking.