DBTB INT Sandy Ryza r

[Music] yeah I'm sandy reza i'm a senior data scientist at clover health um you know it's probably gonna be the same answer as what a lot of people say but it's really sort of thinking when you work as a data scientist you're in your own little world and that means both working on assert like a your own little sets of data your own set of problems but also means like working in your own mindset I've had a how to deal with data right like you sort of cumulate this tribal knowledge of like this is the right way to do things so I like you're like going to conferences and data by the bay so you know pretty pretty good one for this kind of stuff is a lot of like cross-pollination I like going to conferences where you can sort of get you know an understanding of how other people just view the whole enterprise of doing data science I don't think did it so cool and excited I mean I think data is probably always been cool and exciting just now we're in a place for those lots of tools and lots of computing power that allows us to actually do stuff with it yeah I sometimes ruminate on why data science became such a big thing now and not a little while ago but i think it's ultimately just the tools have been able dit [Music] today um so my talk is the talk on the spark time series project it's about how to use spark to analyze time series data I think the you know biggest insight from thinking about that kind of that kind of thing like building a library to analyze data and x rays did in particular is that the layout is extremely important like that's one of the most crucial decisions when you are building that kind of thing how do you lay out the data you know what gets grouped together that enables answering it that affects the data speed at which you're able to answer a certain set of questions so it's like think really critically about the set of questions you want to able to answer and then think about which layout makes sense for that set of questions [Music] there's a lot of advice sort of about what gets gets undersold about being a data scientist so I'm a little bit reluctant to phrase it in terms of that I think engaging with the data you could sort of never go too far with thinking about like what does this data actually mean that like you're never going to be like oh I wish I had spent less time you know trying to understand what's actually going on with this data and also it would have been so much better and then just like sort of a curiosity like my particular way of learning about things is I sort of I read something and I don't really understand it at all and then I read it again I understand little bits and pieces of it you know it's not necessarily like unraveling a thread as much as it is like you know looking at like a Polaroid and having it sort of slowly turn from a blurry image into something with distinct lines and instead like being able to do that like read something and be like it's okay if I don't understand any of this I will understand it later takes a lot of the pressure off and makes it a lot easier to learn [Music]