Summary
The user is exploring whether to use a feature store or an orchestrator with caches for a web application that focuses on exploratory analysis and decision workflow tracking. They seek the best option for on-demand access to both raw and refined data, highlighting the importance of efficient resource utilization and access to well-defined, up-to-date refined data. The response explains that depending on the needs, both options may be necessary. It describes Flyte as a tool for running tasks reproducibly and managing compute infrastructure, particularly for tasks requiring different containers and resources. However, it clarifies that Flyte is not a feature store, which functions more like a database, and suggests that an orchestrator will likely be used to access the feature store consistently.