Elliot Marx

Snowflake has become the backbone for model development and analytical workloads. As production ML systems increasingly depend on signals that change continuously and live outside warehouse tables, traditional feature store architectures built around prematerialized pipelines introduce latency, duplication and training inference drift. In this talk, Chalk's Co-Founder, Elliot Marx, unveils a computation-first feature store architecture where Chalk extends Snowflake into real-time decision systems. By executing Python feature definitions directly against warehouse and live data sources, this ensures that consistency and freshness are properties of the system rather than downstream optimizations.

Real-time machine learning depends on features and data that by definition can’t be pre-computed. Detecting fraud or acute diseases like sepsis requires processing events that emerged seconds ago.