Hi Gianmarco, I've only been speaking with a very small subset of Beam community, so do not purport to represent overall. There is a strong interest in ML by many and certainly by me. No specific 'project' at the moment -- just enthusiasm from bits of ideas from varied exposures that tell me that the two ideas/communities could do well together.
I am a user of Beam (as far as developing it, have only contributed a small bug fix up through now) though ML is an area I'd like to see more of (though, my runners/DPSE are largely Spark and DataFlow on GCP which aren't covered by Samoa) and it seems Beam helps address compute, so working with it frees to focus on the development of Streaming/ML algs. It actually isn't 100% clear to me - at this point - how this will all work, so I have plenty of HW to do. Given what I'm hearing, the best course of action seems to be for me to just start hacking around on something that can leverage Beam and address ML Algs. It doesn't sound like that is an issue from the perspective of animosity for stepping on your territory, or something like that. Should that gain a bit of traction, I will then look forward to figuring out whether that might serve to bring the Beam and Samoa communities together (also, in the meantime would let me get more up to speed with both, and help me appreciate how difficult these problems really are). I do have some dev cycles to devote to this at the moment, so will aim to explore further and see what can be built. Will certainly be in touch with updates! Cheers, Austin On Fri, Aug 3, 2018 at 10:38 AM, Gianmarco De Francisci Morales < [email protected]> wrote: > Dear Austin, > > Thanks for your message. > Indeed, we considered Beam at a certain point, but the project was still > very young and it was not clear where it was headed. > We could definitely re-consider it now that it has gained more traction. > Are you starting to consider having Machine Learning algorithms inside > Beam? > We would be very happy to hear your thoughts and know a bit more about the > project. > > Regarding SAMOA, all the main developers currently do not have much time to > work on it, but we hope this might change in the next few months. > > Cheers, > > > > -- Gianmarco > > On Wed, Aug 1, 2018 at 8:35 PM, Austin Bennett < > [email protected]> > wrote: > > > Hi Samoa Dev, > > > > It has struck some of us in the Apache Beam (https://beam.apache.org) > > community > > that we have complimentary if not overlapping projects in many respects. > > We were discussing yesterday that Beam is trying to handle the ability to > > be cross DPSEe (SAMOA term), though more generally handle the compute on > > top of a number of 'runners' (beam term). Many of us have strong ML > > interests, and around Streaming ML. > > > > We were starting to brainstorm around how these projects could play well > > together. Prior to implementing things, the thought was to reach out to > > learn of Samoa dev's thoughts. It doesn't look like there are ongoing > > project updates, and haven't seen code pushed in awhile. > > > > Has anyone in the Samoa community thought much on this? Are people > > following this mailing list? > > > > Thanks, > > Austin > > >
