A day in life of a data scientist | Data Day Texas | Interview 12 authors/speakers

Key Takeaways

hello it is a Saturday I'm going to data day Texas there are a lot of amazing book authors there including my favorite authors so I'm super excited to go check it out [Music] [Music] foreign delivered the opening keynote the room was packed of people she talked about what is data mesh and how do we get there I enjoyed working now thank you oh thank you it's actually quite hard to choose which talk to go to next since there are so many amazing talks happening at the same time I went to Hannah Nielsen's talk on how math simplifies AI her book essential math for AI is quite a nice read then I went to Hawaii shapira's talk on databases databases are everywhere but they're quite limited how can we get databases to do more [Music] hi hello that's a microphone yes hello good to see you again how are you how are you are you having fun I am a lot of good talks [Music] how's your day it's going well I'm so happy to meet everyone who gave me people's feet in person that like I knew online and meeting people I didn't know in person for or online uh it's going good yeah good to see you again I don't even know if I'm gonna publish anything or not I get so confused every time someone there's so much foreign [Music] [Music] [Music] [Music] yeah so what is meta programming right programs I write programs yeah so how do I do that there's two different ways um in Python uh the the most common tool is called molar it's open source yeah okay and so uh you you can write a Waller program and you sort of specify like what to match on and what to replace things with and so you can use this to automate API migration there's other tools too like there's there's other forms of generated code uh so for example sometimes bindings C plus plus libraries are automatically generated like things like Swig but mostly when people think about meta programming they're thinking more about uh transforming the code do you actually use that in your work I do yeah we use that because we have a large number of people who are using sort of folders or apis and we want to get them on newer versions so that they can take of all of the performance benefits that are in Python 311 because they're on a much older version of python right now we want to get some new version so that you know their jobs are more efficient I say So Meta programming is helpful when your transitioned to different python versions yeah it's so I think um so you you remember like uh two to three right yes that was painful yeah super painful but like a whole bunch of tools to help and they weren't perfect but like they helped right yeah and that's sort of like that's that's the type of meta programming and that's like one of the times one of the more frequent times where people will use it because doing it by hand is just too overwhelming so you you tend to use tools yes that makes sense you have to do stuff like that hi there welcome to day to day in Texas I'm going to be talking about the modern data stack Evolution uh the Journey of different components that make up Monday secondhands change over the last 10 years and why 2023 is going to be the year of the consolidation of the modern data yeah you also have a conference coming up right I do yes I've got a conference in Switzerland called skid skiers in data so it's going to be held in verbier got a lineup of some of the best data influencers and speakers talking about um different different parts of modern data stack and how we can add more business value quicker and easier perfect thank you so much no problem hello Ryan hello could you tell us about your talk yes uh my talk is your laptop is faster than your data warehouse and basically it's a talk about duck DB which is an in process analytics engine it's it's going to be a really fun talk showing lots of different queries capabilities of ductdb a little bit of the internals how duck DB Works under the covers that type of thing perfect where is your Duck shirt where is my duck suit uh we're not at coalesce right now this is day-to-day Texas the duck Suits come out at coalesce um okay I love that I really like your Duck shirt um what else do you have a book are you ready I do have a book uh but my book is unrelated to data so I I consider myself a geek in two different areas one is data and then the other side is authentication authorization so I have an O'Reilly book on on oauth but it's many years old and I haven't gotten back to updating it so uh because I've been really focused on data right now with with duckdb and and mother duck is a cloud-based analytics platform that's building on top of duct TV but in the past I worked on Google bigquery and worked at neo4j and worked at databricks perfect so when is our next book when is my next book wow um I don't know maybe you should ask the O'Reilly folks they were pinging me um uh no I mean I just uh I just co-founded this startup uh with a bunch of other folks um so I've been a bit busy uh and I suspect to be busy for a bit so uh eventually I want to write a book around developer relations which is oh I love it uh what I've been doing for for many years and and it works differently inside of every different company so uh eventually I want to write a book on that and uh I started to convince the O'Reilly folks but uh thank you so much Ryan thank you hi Matt how's it going good how are you good it was great to meet you officially at this conference and thanks for hosting the reading group before as well where we met the first time thank you so much could you tell us about your talk today yeah um so we're going to be talking about the we're going to call it the state of data basically and it's actually mostly going to be more of a town hall because there are so many amazing people here at this conference who've done all kinds of fantastic things so we kind of want to just discuss their opinions on where the industry is going right now what the trends are how data professionals should cope with you know rapidly changing kind of chaotic economy that we're facing right now and also how to think about training new data professionals because I think that's a huge issue that we're all kind of facing like how do you actually get people into this profession and what's the process for doing that perfect and thank you so much for signing the book for me yeah of course can you tell us briefly about why should people read a book especially to the scientists yeah I'll give you a little bit of context on this um so I co-authored this book fundamentals of data engineering with Joe Reese and it's available from O'Reilly O'Reilly media and so you can buy it on Amazon or get a digital copy from O'Reilly's website and our goal really was to kind of tie all the pieces together so there are a lot of fantastic instruction manuals for different Technologies you can use but the problem is there's this huge proliferation of Technologies and so our goal was to like kind of put those all into context and show people how to stitch them together certainly data Engineers who are just entering the field or who are growing in their careers but also data scientists it's not intended to like teach data scientists to be data Engineers but rather to help them to understand what the pieces are so they can be more effective in communicating with data engineers and kind of giving input into architectures to do their jobs awesome thank thank you so much of course thanks for the discussion hello Andy how are you I'm good thanks a lot so yeah and you have given it up already right I did I was talk cool very cool actually I love you know the energy in this even so um the type of question so also the setup at the beginning which was interesting you know you have to do it all by yourself which is great so you put your microphone alone thing in the pocket you bring everything back up it's cool so yeah what was their talk about my talk was about automating the data monitoring right and how this automation of data monitoring can lead to a data absolutility uh thinking right so yeah that was my talk which tool are you using which tool which tool am I using for what monitoring data uh monitoring so I'm using my platform right uh the cancer platform and uh and the way it works is that we are the platform is decoupling two things here is the agents and the agents are these kind of instrumentation of the data tools themselves think about five transport whatever the agents are generating information to the cancer platform which is the second component where the the monitoring and obserability is applied awesome do you have a book will I will have a book now I'm writing of this fundamentals of Delux already book at a rally yes uh currently trying to finish the fifth chapter so still two to go afterwards and then yeah hopefully by September it will be on the shelves this December uh this September September yeah so amazing looking forward to reading a book yeah thanks me too thank you yeah do you like to follow me on my YouTube that's what I'm trying to do now so let's let's find you so see ya and then G [Music] y uh y not w sorry and G yes that's it yes Anaconda is that you there you go either way it works there you go and then subscribe there you go right there let's search Sophia young and you will see me on YouTube okay I need to watch the watches now so when you have your new book I want to interview you cool for a new book awesome Yes um do you want to get lunch yes let's go let's go live okay so far I love it [Music] okay foreign [Music] area and are interested in publishing um I cover mainly data analytics nice so whoever wants to write about contact you yes please yeah it's a lot of fun I love your shirt thank you yeah that's funny it's a bit yeah so can you tell us about your talk this afternoon well I think you're gonna have to stay tuned but the uh depending who shows up we might give a talk about fundamentals of data engineering or if people are lingering around uh for happy hour we're going to turn into a town hall so I don't know which talk we're giving yet I think we're pretty flexible but it's gonna be fun either way that's amazing you have two versions of the talk yeah two and a half actually yeah two and up oh my God yeah always be prepared next question would you tell us about your book and why should data scientists read it uh yeah so fundamentals of data engineering um is a book written by two recovering data scientists that's Matt and me we wrote the book in large part to I would say help people really fill in the gaps of foundational knowledge of data engineering so if that's something that interests you then you should read the book thank you so much thank you very much I appreciate your time could you tell us a little bit about your books and your talk today uh of course so my name is Kathy Chang I'm a co-author of how to lead in data science it's a practical guide for data professionals across different parts of their careers and today we're doing two talks we just did one on 10 mistakes to avoid for the data professionals or data scientists and the second talk is coming up in the afternoon at 2 50 p.m it's on how to what what to do when there's so much to do for the overwhelmed data professionals hey Dad yeah the first one we talked about all the things you can do to be able to be more rigorous in your data science practice there are many things that people are doing quite well but often as the your practice matures there are a lot of complications that deals with the next steps in getting the prevalence of usage of data in your organization and how to make sure it's scalable and unpaintable over time so we suggested a lot of different things and we want to be rigorous ourselves and in the afternoon the talk is about how do you look at all these opportunities around you and to be able to prioritize them and make sure that the organization is on the path where it's doing the projects and the initiatives with the highest Roi and be able to get the most accomplishment in a very short amount of time I'm part of the theme today is to help those who may be in the market for for a new opportunity given the recent or reorganizations in across different companies as well as some of the layoff activities that's taking place how do people take this time to build a stronger more powerful narrative to makes themselves more attractive in the marketplace as well as in addition to themselves also their teams as well love it looking forward to the talk thank you thank you now could you tell us briefly about your book why should all data scientists and data analysts read it so I think I've worked in data science organizations large and small and it's a Monumental task to be able to see an organization that's making a lot of impact in the marketplace but managing a team or as part of a team that's quite small often less than 10 people fewer than 10 people so in those situations many of us have to take on multiple roles at different levels of influence sometimes team leaders have to be talking to Executives on some days of the week and the individual contributors in other days of the week now when they're taking on those responsibilities and providing those influences they're actually very different thought processes and different levels of things thinking that's required and often those are quite hard to tease out and when we don't do those row mental role switchings when we talk to different stakeholders in the company sometimes it leads to confusion and setting up different expectations that we cannot meet so this book really lays out the expectations and degree of influence one can take on through those weeks that challenging weeks where we're doing a lot and to be able to do those mind switches and be able to Target the audience we are facing either whether it's Executives or cross-functional Partners or our team members to make sure we're setting up the organization and the culture for the organization for success so how do lead in data science again it's a practical guide so to make it more consumable we organize it in a way that has four part four main parts um that each part focuses on on a particular level of role so data Science Tech lead manager director and executive and part five focuses on applying the analytical rigor to the process of developing your career and to make it easy to use we enclose case studies there are seven of them that we provide a situation and useful Concepts the strength opportunity and support requests for that individual for that Persona aligned with the sections of book that you can jump into so if you find oh that's me Jennifer has exactly what I'm facing so go to these sections and read up on them and there are also 101 gem insights so if you don't have time to read any part of the book in whole open up the book in any section find a next gem that may attract your attention and read around it so we took that into consideration while writing the book because we we know everybody's busy but everybody has a willingness to learn so let's make it easy for them to do so perfect thank you hi Hi how are you doing fantastic what about yourself what was it talk today it was a lot of fun I spoke about uh how can we build architectures for the next 10 years of growth uh focusing on data discoverability so especially integrating Discovery Platforms in order to understand where data is who produced it what should we do with it and so on and also the fact that data is transient today in this world and so we should start looking into git for data and data version controls and how these tools enables us to solve a lot of critical problems in the space of data and also connects to the rest of the data platform which is really critical for us because at the end of the day we need our systems to integrate really really well in order to be successful building data products I know you're working on a friend you tell us a little bit about yourself absolutely so it's actually it should be printed next week and they might even have copies here I need to take a look uh it's on scaling machine learning with spark and also going beyond that like how can we build a data system that's going to support our efforts in building scalable machine learning um and also going beyond that because what we realize that's well spark is a fantastic tool when you have a lot of mlips algorithms Implement them into it you also probably need to leverage other tools for deep learning such as high torch tensorflow and so on and so understanding how the distributed work uh workload works there is really critical for us as we build those systems and lastly deployment like what are the different deployment patterns for machine learning observation and they addressed model drift yeah all this good stuff look forward to reading it thank you thank you appreciate it how are you doing doing well so you just had your talk today uh what was I talk about um so we talked about data contracts accountable data quality how you implement data contracts what they are the value the problems they solve so for people who have never heard of a data contract what is data contract the data contract is a agreement between a data producer and a data consumer we're a data producer is someone that owns a database or some other Upstream data source and a data consumer is someone Downstream that uses the data to do something interesting in machine learning or analytics or whatever it might be and the contract is that agreement that is enforced programmatically so it's a way of moving data quality from the right on the consumer ownership side to the left to the producer ownership side where can people learn more so you can learn more on my sub stack which is data products just dataproducts.substack or my slack Channel data quality.camp slash slack we talk about data contracts all the time or you can follow up with me on LinkedIn I do a lot of writing there pretty much every day perfect I just joined your slide this week okay thank you for joining that's right maybe or start a new company maybe maybe I haven't put pen to paper yet but I think that there is a lot of opportunity to do that so just trying to find the best way to do it and take my time nice looking forward to it thank you thank you Chad okay this is Patrick okay he wrote the book managing management on kubernetes amazingly it's only that much information oh wow because kubernetes is so easy no it's not all right nice signature on the bus this is now worth three or four more cents I'm gonna say three more million dollars so it's funny you have the error books well it's funny I got an email from Jill and she's like there's been an error and I'm like oh my God what's happening yeah and she's like there is an error in the printing and I'm like but you know they just printed a bunch of these books we're very mortified and all this stuff like what was it we forgot an eye oh that doesn't seem so bad no now it's a Collector's Edition that's right that's right but that's on the first page it's on the first page and then I love her like oh that'll be hilarious yeah but that would have made it more interesting right but um I figure with with kubernetes a couple of Errors we could just absorb it it's kind of the theme of the book yeah fault tolerant yeah there you go yeah exactly yes yeah do you want me to sign all of these sure if you sign any number of them I'll sign them on could you sign one for me absolutely it's a showcase yeah you got sophh yeah s-o-p-h-i-a I just let all the people yay thank you you're welcome [Music] foreign Robinson I'm the lead data scientist at graphable well I specialize in all things uh rap machine learning and graph analytics um so today I'm giving a talk on introduction to graph data science for python developers I'm super excited to be here what kind of tools are you using uh a lot of python um so integrating python with all kinds of other libraries so um deep graph Library seller graph as well as graph database Library so integrating things with graph data science with neo4j or tiger graphics so thank you so much I went to several great talks in the afternoon I really like we don't dance talk on artistic processing the untapping power of data visualization where he combined data with Arts so many inspiring visual reputations of data Ryan Boyd talked about duck DB they believe that big data is dead because of the definition of big data has changed over the past years and people actually don't use Big Data those queries query small sizes of data and finally Joe and Matt's closing keynote is amazing the town hall where people discuss the data profession training and education and more [Music] thank you [Music] thank you foreign [Music]