Christian Kleinerman, Snowflake | Snowflake Summit 2023
summary
Claims a single engine runs SQL, dataframes and Java with no performance tradeoffs and shows a path from parquet files to managed Snowflake tables via Iceberg unified tables that support unmanaged read only and upgradeable managed modes. Announces Snowpark Container Services plus a native app framework and app store for distribution and monetization, with git sync, logging and tracing APIs. Demonstrations combine Document AI, embeddings and Neeva style retrieval to keep data and governance inline, customers report being mind blown by the possibilities.
AI-written summary · about
chapters (14)
- 0:00Snowflake Summit overview
- 1:33Unified data engine strategy
- 2:50Breaking down data silos
- 3:46Iceberg tables introduction
- 4:46Standardizing application data
- 6:19Container service use cases
- 7:25Data engineering and LLMs
- 9:14Governance in gen AI
- 10:38Secure application validation
- 11:57Neva and trusted AI
- 14:01Vector search integration
- 15:32Platform and tool strategy
- 17:07Future architectural limits
- 18:37Customer feedback and outlook
