Interested to get a feel for what's coming down the pipe. In the exploratory phase of assessing if the technology appropriate to adopt for a new project we're working on.

Your other question in terms of what is coming down the pipe we have better answers for. We are still cohering our concrete roadmap for a next major release (tentatively codenamed Santa Baby 🎅🎅🎅). We have a few themes though: 1. Cloud-Native execution: So far we have focused on our programming model mostly with only modest support for DevOps. This will change very soon. We will have a out-of-box k8s deployment ready to go along with on-demand ephemeral compute for runs. We are still working on the exact shape of this. 2. Ability to queue runs for cloud execution. This directly applies to 2, but will be have new APIs for GraphQL and easy-to-use tools for enqueuing runs. This is enabled by number 1 and enables number 3. 3. Explicit support for backfills. This is a critical activity for managing data asset generation. Now that we have native scheduling abstractions, we will be building what we think is a clean backfill abstraction with a more generic/abstraction partitioning mechanism that we will think will be useful in other contexts as well. 4. The ability to schedule ad-hoc computations in the cloud using is also very useful for ML workflows as well. So at the end of this burst of work we think we can support end-to-end data engineering and data science workflows, because you can express your DAGs and dependencies in a single unified system, rather than having to artificially silo them because of limitations or overspecialization of current tools. Hope that helps!