Willem Pienaar

> I believe a key developer experience component is the seamless consumption of a precomputed feature along side of a derived feature, both serving and batch. > Requiring the derived...

> Is it possible to use TensorFlow Transform, Apache Beam or other frameworks to generate features? Yea this is certainly possible. I think the question is just what the developer...

> One of the common use cases that we have internally is derived features that are based on the output of other ML models. > In addition, we have 2nd...

> Is this still somewhere in the Feast roadmap (and if so is there somewhere I can track the progress)? It's a feature I would love to see, and am...

As an update here, we are currently working on an RFC for on-demand transformations.

Thanks for raising this @weitian! Would you mind providing a way for us to reproduce the problem? What led to this comparison being needed?

Just swinging back here @dmatrix. We'd love to support the sources you've laid out above. I think the biggest challenge from our side is just how many storage implementations we...

> [We will be using kedro as "glue" ... we run workflows in argo-workflow, and kedro can build them for us. Obviously Feast will help us out with our feature...

> > training_df = store.get_historical_features( > > left_table="drivers_activity", > > feature_refs = [ > > 'drivers_activity:trips_today' > > 'drivers_activity:rating' > > ], > > ) > > Does this mean...

> Should there also be an option to "keep latest" only, when used in conjunction with the time range filtering > Otherwise its more than possible that the underlying entity...