What is a big data scientist?
Transcript
hey how's it going as he thousand from the CTO visor calm with today CTO daily dose for October 23rd 2017 we're gonna talk a I infrastructure in AI and what is a data scientist got this question on Twitter last night actually I didn't post a question another person post it as a Twitter survey and I thought it'd be a great topic for today he's CTO daily dose so we're going to start at the infrastructure I'm not an AI machine learning expert but I think I know
enough about the topic at least introduced a high-level concept of what is a data scientist know think set definition for a data scientist but I think walking through the concept of data science will help determine if that's a correct career path or if you're interviewing resources for for project you know how to classify the concept of a data scientist so let's start at the infrastructure we have data first you have to have data without data there is no science there's nothing to analyze that data can
be anything from huge object repositories or small object repositories they can be structured unstructured data inside of a sequel database Oracle database or NFS file that's really not the major concern but you have a large data set and you connect that data set usually via some high-speed connectivity to the compute and specifically in compute we're talking about TP use and in GPUs so something like Google's open source tensor flow which uses specifically built machines or processors to process machine data or GPUs systems from folks such
as Nvidia and their kyudo platform that's used to process that data so you have infrastructure architects and infrastructure engineers that build this high-speed data network for consuming the data so consumption of the data is something that we are pretty solid with from a traditional enterprise architecture enterprise engineering discipline that I don't really call that data science that's typically what we do in the enterprise however the expertise needed to take advantage of the tipi use in a GPUs I think starts to bleed into the concept of
data science or at least supporting the data science so creating the algorithms and models to be used in machine learning and AI is where the true value lies and in order to make that happen there needs to be api's available to that compute and storage to create the models and this API layer and I think is where a lot of the magic happens this is where we will get our you know our Python developers our gold developers our application developers who are actually creating and implementing
models our basically math models to take advantage of the data set and an ally and and perform analysis a second role that this API model could be the folks writing apps and services that enable modeling example is AWS has announced a ai platform in which you can consume that AI platform one of the the conversations I got in with Michael Bushwood is the at that level it can be as simple as using Excel or it's complicated as creating oracle DBA of what oracle DBA programmer would
do and even more complex than that so data scientists pretty much in my opinion play in between these two levels their level in which they're actually creating models scientific models to consume the data and there or there creating applications or taking advantage of applications which are there creating new models in order to mine data that's my high level definition I'd love to hear feedback or pushback if you think you have a different model or definition of what a data scientist is I'd love to hear it
hear it follow me on twitter at CTO advisor or on the web the CTO advisor comm or you know what LinkedIn a great place to have a conversation let's talk and I'd love to hear your feedback until then talk to you guys next CTO daily dose
