Development Team discussed on Software Engineering Daily

Automatic TRANSCRIPT

Have you felt a need to aggressively higher machine learning developers or more data scientists how are you thinking about the development of your data teams so i think proportionally wise we actually have a pretty good number of dada scientists so for us in a group we've got sort of three to full different data scientists and what happens is they kind of end up sitting down and working closely with a product team so if we have a product where we think that needs to be some sort of machine learning or some sort of dada science kind of project this collaborative dynamic teaming is kind of a characteristic of how we kind of work and so they'll sit down with the team and then agree on exactly what is that they'd like to maybe do what maybe approach training modal would actually make sense and then trait actually production ising that with the different development team so it was quite well for us it doesn't really make sense have thought assigns everywhere not yet but it's really very focused on where we think the most practical uses but we have very good sort of collaboration that means that we know people made a web or leeann as to whether this thing would be a good place to have some data scientists look bit closer to help us improve the experience so it's exactly what what i was thinking about is you've still got probably a lot of greenfield just like feature development application basic applic core application development product ration to go before the machine learning side of things can really be be higher leverage than that because the machine learning side of things is more of optimization over products that have already been proven in developed significant data sets i mean at the same time we have a very large base of more than a million people and we've processed on a billion euros a month and so this actually put a little daughter in the i mean we use it in some simple examples so for example when you make a transaction sort of transaction list when his we have some classifies a try to classify and give a hashtag as to what it is this ends up being used in terms of statistics that we demonstrate as to you'll.

Coming up next