Community Spring Cleaning week is here! Join your fellow Maveryx in digging through your old posts and marking comments on them as solved. Learn more here!

Alter Everything

A podcast about data science and analytics culture.
Episode Guide

Interested in a specific topic or guest? Check out the guide for a list of all our episodes!

VIEW NOW
MaddieJ
Alteryx Alumni (Retired)

We're joined by Raf Olbert, Data Scientist at Asda, for a chat about the rise in the volume of data, the agile manifesto, and our shared passion: Aviation.

 

 


Panelists

Brian Oblinger - @BrianO, LinkedIn, Twitter
Raf Olbert - @rafalolbert, LinkedIn


Topics


Community Picks

Brian:

Raf:


Transcript

 

Episode Transcription

BRIAN 00:13 

[music] Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Brian Oblinger, and I'll be your host. We're joined by Raf Olbert, data scientist at Asda for a chat about the rise in volume of data, the agile manifesto, and our shared passion, aviation. Let's get into it. [music} 

BRIAN 00:44 

All right. We're here with Raf. Raf, how are you today? 

RAF 00:46 

I'm great. Thank you, Brian. Thanks for having me on the podcast. You've got such an amazing track record of some great guests, just great to be a part of this. I'm a big fan of the pod. I'm an Alteryx addict, as we'll soon learn. And, in general, I like the podcast medium. It's fantastic with so many commute hours of the day, [inaudible] hours, and hopefully, some exercise, it's just maybe ad Infinium medium to task. But it would be such a missed opportunity not to have this supplement for your audio education, so overall, just massive thank you, and thank you for creating this amazing podcast. Thanks. 

BRIAN 01:41 

Wow. That's nice. I appreciate that. And thank you for being on. I'm really excited about our conversation today. We've got a lot of good stuff to get into. I think it's going to be awesome. So let's go ahead and get started. So why don't you tell us about you, kind of how did you get into data, how did you get into analytics, and kind of what are you working on these days, what's kind of interesting in your world? 

RAF 02:04 

Absolutely. So first of all, my name is Rafa Olbert, but I go by Raf, just to make things easy and snappy. The actual spelling is very close, surprisingly close to RAF, Royal Air Force, and aviation is another passion of mine. But, in my current role, I'm a data scientist for the e-commerce side of the business, working out Asda UK, so I'm based out of Leeds. As an organization, we are part of Walmart. So with this global scale, we are the largest organization and enterprise in the world, revenue and people-wise. Where I can share an interesting fact with you, we are also the largest deployment of Alteryx globally. That's both the back and server side of things. And the number of the Alteryx designer licenses we've got-- I'm not able to quote any specific numbers, but they are impressive, trust me. 

RAF 03:20 

With my current role, I look after data optimizing, as well as the strategic goal of building the internal competency for AI and ML developments. Before Asda, I worked within an aviation enterprise for a couple of years, looked after the wider IT space. Set up the internal intranet, also looked after the enterprise resource planning, implementation, as well as-- I was very lucky to get my hands dirty with some really technical and logistics side of things. I got to look after airworthiness and resource utilization. Probably the last thing, just backtracking from my current and previous role, my educational background is-- I've got two master's degrees, both in [inaudible] and management, where I specialize in enterprise. So surprise, surprise, I'm working for the largest one out there. And I also majored in computer science, where I specialize in database design, so that's top level. 

BRIAN 04:56 

Yeah. Just those things, nothing else, right? 

RAF 04:59 

Yeah. Yeah. No. I'm also married with five kids, so just top if off with this. I don't know if it's made up or we found it somewhere with my wife, but we live by this real simple rule, "The more you do, the more you do." So it's just trying to keep ourselves occupied round the clock. So yeah. It's lots of fun, lots of fun. 

BRIAN 05:28 

Well, yikes. That's a lot of stuff. That's amazing. And that brings us to our first topic. So you've got a lot going on, clearly. You've got five kids. You've got this fast-paced world you're living in. Tell me a little bit about how do you navigate that. So how do you adapt with that kind of pace? And how are you keeping up? What kind of efficiency life hacks, if you will, are you employing to make that happen? 

RAF 05:55 

Absolutely. So that's probably an ideal place to introduce Alteryx, and where it's kind of helped us over the last couple of years. I think we operate pretty large data, or Alteryx sustains with over 120 data [inaudible] running on a daily basis over 20 apps. And the actual background was that we came from the XL solution. I was coding [inaudible] left and right. The actual very subjective for my employments into the previous role, which was the BI lead developer was to automate everything using XL. So we had pretty good routines. Some that did some really crazy stuff, just web scraping. But it was probably three, four years ago when we came across Alteryx. And it was love from the first sight. We never looked back. You can probably imagine, it was pretty hard, where all the previous experience and knowledge acquired that almost defined my role and my kind of like position and the success. We had to give it all up in favor of Alteryx. But we never looked back. 

RAF 07:31 

So, we really embraced this self-serve, kind of like private [inaudible] and probably the frontline. As you can probably imagine, the pace, especially on the e-commerce side of things, is just unbelievable. And I like to reference its attention rise. So with all those ideas, and the aim of being first to market, and all those competing bright ideas that you would otherwise just spend ages refining and then comparing, just trying to sense what value they may bring, Alteryx really allows us to do quick and dirty, to connect with every moving database that's out there. And the actual data landscape is pretty rich. So we cover all different traditional RDBMS technologies that we can probably cover in the pod later. But the pace is just unbelievable. So with this attention rise, we have to be able to quickly sample the data, just blend across all different technologies, and just arrive at some meaningful results, discovering the insight and almost projecting the value gain behind those projects. 

RAF 09:08 

So Alteryx has been transformational for us. And even now, with my current role, where this kind of things that we do is slightly beyond the kind of like operational capabilities of Alteryx. We operate over big data using spike. Alteryx is still part of those design flows, either on the first leg that they input data, or actually covering the last leg of the analytical side of things. So yeah. It's been a great ride, and we continue to embrace Alteryx. It's almost like this reference point, or a common denominator that brings all these different perspectives of technological overview and the stakeholder engagement, and playing the Alteryx pipeline just makes things so obvious and so simple to the fine level of detail of controlling the records that count. So yeah. It's been great, and it continues to be an amazing ride with Alteryx on board. 

BRIAN 10:35 

I'm just curious, you as kind of an individual, from an efficiency standpoint, it's one of the things we talk about a lot on the show with guests. What are the things-- so you've got a bunch of kids running around. You've got all this work stuff going around. How do you organize that for yourself? Are there certain systems that you've built over time, other than, you said you do more, and then you do more? Are there certain systems or efficiency things that you've done? How do you prioritize all this stuff and keep it all straight to make sure you're delivering, both on the personal at-home front, as well as at work? 

RAF 11:13 

Yeah. I think there are those key priorities and almost perspectives that I like to introduce into how and what I actually do, both work-wise and outside. And the very first one, it's one that probably is not stacking very well against the most recent key buzzwords, like Agile or [inaudible], Scrum. And trust me, we take all the boxes against all of these. We all praise-- 

BRIAN 11:49 

This is the part where you start telling people how you organized your whole life on blockchain. 

RAF 11:55 

No. I was really in the mindset of blockchain a couple of months ago, surrounded myself with all those inspirations. There are some great titles on Audible and some great podcasts. And I thought maybe we can find a slot for this. I've got a revolutionary $1 billion idea that I can sell to you. 

BRIAN 12:24 

Make sure you get the patent on that before we throw it out on the episode here, I guess. 

RAF 12:27 

Yeah. Yeah. Well, cover it just very top level. So we can send the integrated Alteryx with blockchain. But no, just kidding there. I think the one keyword that I was going to reference is entrepreneurship. Again, it may not stack very well against all these exciting new kids on the block. But it's those kind of like reference point that goes back to my upbringing and early days. I was quite fortunate to be brought up in this periods, where I don't fear just giving it a go, trying to-- every entity in the kind of like wider enterprise space, every project, every database, there is a small, narrow team behind. And Alteryx being this kind of almost the language of communication, where it'll probably sound from my accent that English is not my first language. And going back in time, I remember one of the first key phrases that I learned was, "Excuse me, to speak English." 

RAF 13:50 

So I could almost sweep this around and say, "Excuse me, do you speak Alteryx?" Where minds really connect when you bring this reference point, and this entrepreneurial spirit, just being curious and trying things and owning things, as well. With Alteryx pipelines, we like to operate these kind of like [inaudible], these tile process ID connectivity, again, connecting with so many values, different data points ranging from [inaudible] data to IBM mainframe, DB2, all the way to some-- and lots of different flavors of Oracle and now Hadoop Hive, some [inaudible]. So just having this tool to accommodate this bravery and curiosity of something that they turn, and creating some data blends that have never happened before. Some of those data pipelines, they are really first in the history. And the ISD teams are actually very surprised. We are the go-to team for the lots of things, where Alteryx was the horsepower of those advancements. So yeah, I believe entrepreneurship is one of them. We can probably arrive at a few others. 

BRIAN 15:28 

Yeah. It's interesting. I mean, I think-- obviously, the definition of entrepreneurship has inherently starting a business kind of in the definition. But I think what we're really finding is that people that are really successful, even inside of really big companies, have that type of spirit and that kind of mindset. And it is possible to bring that even to a very large business, where you say, "Hey. Here's something we're not doing today." And maybe we should set up an entirely new department or cross-functional activity, or whatever it is, and be that person who can be the change agent, and the one who's building things. And I think that's where you kind of hear people talk about sometimes, especially in Silicon Valley, you hear them say, "Hey. We're a startup within a company," right, or, "a startup within a startup." And so I think it's a really important skill. It's something people should be thinking about. I know that not everybody's comfortable at the end of the day kind of sticking their neck out and putting their reputation or their career on the line for something like that. But there's obviously some really great things that have been done, and will be many great things that will be done in organizations big and small with people who have that mental model. 

RAF 16:39 

Absolutely. I totally agree. There's another angle to this. I think so. I came across this, just starting the product management space and specifically the agile ways of working. And I really like to think of Alteryx as a tool that tackles both, project management as well as product management, where the first one is really very well-defined, in terms of some boundaries like time scale, that the scaleable functionality that you expect to be delivered behind the project and the stakeholder group. For us, with Alteryx, while we deploy solutions, they are really lively entities that are set out to tackle some problems and to thrive inside. But the point is that I assumed that most projects, at the point when they are being delivered and run, and are deployed, QA'ed and go live, they are already heavily off what our thinking and expectations are at this point. And it's this mindset of working within the Alteryx kind of us, where you almost defy the very light requirements, and almost thrive inside a product kind of like empirical discovery and exercise, where with the standard models, this just wouldn't be possible. So with Alteryx, we tend to own those products past the project space, and we evolve them and make them relevant. And sometimes, they go away, and they are replaced by some new, shiny tools. But certainly, Alteryx has helped us to tackle both angles. 

BRIAN 18:57 

So, in the spirit of learning, actually, if you don't mind, if you would indulge me, one thing I'd love to do here is-- when you were talking there for a moment, you said a bunch of different, kind of emerging technologies and things that I think people may have heard about, but they maybe don't know exactly what it is. And I'm finding, the more conversations we have on this podcast about different things, I hear, and we hear on the back end of this people say, "Oh. Yeah. I'm learning a lot." And so I want to make sure that we're bringing that to the audience. So if you don't mind-- and I'm not looking for the perfect definition because I think everybody has slightly different or mildly different definitions. But if we could just go through a couple of things real quick. And I'd love to hear how you would define it or describe it. Will you play that game with me for a few minutes here? 

RAF 19:45 

Yeah. Let's do this. 

BRIAN 19:47 

Okay. All right. All right. Thank you. I don't want to put you on the spot, so let's talk about-- so you mentioned data lakes. And these have been around for a while. In your mind, how would you describe a data lake to someone that's never heard of it before? 

RAF 20:03 

Okay. So we, as a e-commerce business, we are very fortunate to have been surrounded by, not just mountains but - I always like to say this - not just mountains, but ranges of mountains of data. So throughout the years, I've connected with colleagues and teams that were just not as fortunate, and I was very surprised. But just following on with this great depth and exposure to data-- I probably will not be able to disclose very high-level detail, but as Walmart, we decided with all these regional markets, like UK, China, that we are going, first of all, to the cloud. Same data puzzles that we work with, they simply are truly big data. So I like to reference this three V model of high velocity and high variety and high volume. So it's not just the simple size of the data, but it's the kind of like ongoing pace and variety, so kind of like standard, kind of like tabular data, and maybe some log data or video recordings or image data. So a data lake is-- in terms of how we operate this space, is a Hadoop base. So Hadoop file system is the basic premise of the technology. And it's based on the cloud, so we collaborate both with Google and Microsoft. 

RAF 22:12 

And it's almost this no-limitation space, were we can find different data points. We bring them in [inaudible] so any third-party integrations, any data that we don't have currently materializing against our other technologies, as well as replicating those technologies. We bring them to this one space. We can find certain rules behind access of different, maybe more sensitive data. And an example of this is Clickstream [inaudible]. So with the e-commerce website platform, we collect every single interaction the customer is making on the website. This covers all the search terms, or the add to carts he makes, just to enhance the experience and be able to measure the success rate of different projects. So this specific data point, with this three V description, accounts for over 800 columns and weighs over 5 GB worth of data for just one hour. And we collect this through our Adobe Integration every one hour. So it's simply beyond any traditional IDBMS storage technology, to almost handle the volume. And then to run any sort of analytics. So with data lake, you almost don't only tackle the storage element, which is the entry point, but you almost enable the compute of that whole. So thinking of Alteryx, just sampling some data from the data lake, we operate some script rise analytics using spike to actually cover longer time periods. So the key features are a cloud serverless, no storage limitations, ELT, and storage and compute to be almost treated as this bundle of combo that actually can take things forward. 

BRIAN 24:37 

Got it. So in layman's terms, it would be-- is it fair to say that when we hear about big data, which we hear a lot about, or we've heard a lot about several years ago, is it fair to say that data lakes is sort of an all-encompassing term that actually is comprised of a bunch of different components that are the technological enablement of big data. Is that a fair summary? 

RAF 25:02 

Absolutely. Absolutely. That's spot on. 

BRIAN 25:04 

Okay. All right. So you mentioned NLP, or Natural Language Processing, what's that all about? 

RAF 25:11 

Okay. So we deployed a very interesting project a couple of months ago, where we collect customer feedback on the side through this third-party integration. We used to just collect the feedback for the previous day and assess different topics. So that the NLP processing use-case in this is topic modeling. So it's identifying the key theme, the key meaning behind the feedback submissions, so primarily some problem areas like the checkout is slow. Or the card payment stopped working. Or there is some missing product. So we used to manually tag these, and just based on some influx of some problem-type over day on day basis, or trending over the weekend, we would trace this into internal support systems like Jira. Or Jira, I think it's another pronunciation. So all of this was purely manual. So we still had all those great systems. We had supports from the technical teams to [inaudible] to this and maybe identify some threads and patterns of problems behind the scenes. So with the use of Alteryx and this data science platform that we are internally developing, we API the data size. That's where the Alteryx comes in. So we do all this kind of like authentication, two or three, kind of like handshake. It's very intricate. But it's absolutely amazing how Alteryx can play this power agent, that otherwise you would need to code using KRNL. So the download tool has been just amazing with this use case and multiple others. 

RAF 27:38 

We then pass on the back-end data to the natural language processing engine that we developed ourselves. So it's [inaudible] preprocessing, just cleaning of the data. In machine learning, everything needs to be a number. So we vectorize all those comments. We use multiple different approaches that we fine-tune using what's called hyperparameter tuning. So we use [grid search?] where multiple algorithms take multiple parameters and with this great approach you can scan across all these different variations and identify those that perform the best. And the best part behind our setup there, it's a supervised learning situation. So we had previous data that was manually assigned all different topics or problem types. So we could almost [inaudible] some port. The usual split is 70 to 30 or 80 to 20. We may use some additional holdout set for further evaluation. 

RAF 29:02 

Because the hyperparameter tuning is actually evaluating on this kind of like secondary subset of data. So it's almost queueing your understanding. So you may provide this holdout set for this kind of like final, almost imitating the real-world scenario, where you're playing new comments and you still evaluate the accuracy. So Alteryx then comes in again at the end of the cycle, where we integrate using APIs with Jira and ServiceNow. So the full cycle just operates behind the scenes. So it's connecting our customers with engineers, being able to work on those issues around the clock. And it simply wouldn't be without Alteryx that we are able to pull this one. It's been up and running for a couple of months. And we are now planning again-- speaking to this product mindset, we are planning to enhance this product by additional data sources, Twitter feeds and multiple other surveys. So it will grow in scale, and it's been great. 

BRIAN 30:27 

Yeah. So okay. So let me take a stab at this, again, from sort of the layman, like if my mom was listening to this. Hi, mom. So I think one of the hardest things for people to understand when we talk about AI, machine learning, and then you just talked about what natural language processing is, is how they nest together. It's a nesting Russian doll, right? So let me take a stab at this. So artificial intelligence is the term that is the umbrella term for all things where computers are trying to replicate or enhance data and information in a way, either like a human would, or in ways that are better than a human would, right? That's what artificial intelligence is, right? 

RAF 31:18 

Absolutely. 

BRIAN 31:19 

Am I on the right track? Okay. 

RAF 31:19 

Amazing. 

BRIAN 31:20 

Great. So then machine learning is one level down on that rung, and machine learning is a type of artificial intelligence in which we build models, statistical models, using data that basically take existing data you have, and we train the model to say, "Hey. If these conditions are met, or these conditions aren't met, or you discover a new pattern, then this is the answer." And what makes them the learning part of the machine learning is, as new data is generated and fed back into the model, it can then take it against those new conditions and those new learning patterns and say-- maybe the model actually changes to say, "Well, we thought it was going to be like this, but it turns out, with all the data and what we're seeing, the model is now looking at things slightly different or in a wholly different way." That's machine learning, correct? 

RAF 32:14 

Absolutely correct. 

BRIAN 32:16 

Okay. And then, natural language processing would be a subset of machine learning, which is simply to say, "We have this unstructured data--" which is basically what we're doing right now, right? We're talking into microphones, and it's being recorded on our computers. And you can point a machine learning model at that data, and it can listen to what I'm saying right now, in my voice. And it can parse that into words or phrases or sentences or even intent or other things, maybe even sentiment analysis or other types of language processing, to understand, what did Brian say, maybe why did he say it, what kind of accent does Raf have and why, right, depending on how sophisticated our models are. And it can learn a bunch of stuff about that and take that unstructured data and now make it structured in something that can be perhaps fed into another model somewhere else to train something or learn something and provide us more insights. 

RAF 33:17 

Absolutely. I think there are two different schools where NLP is sometimes promoted alongside machine learning under the AI umbrella. But I really like to reference the AI kind of like-- there is this key term AGI, so Artificial General Intelligence, that is the kind of like ultimate goal, the scary thing where machines take over. Whereas the ML and NLP side of thing, this is more looking down the road, if it needs-- it's nothing new. It's recycling the knowledge and expertise we had for probably a good 10, 20 years in the statistics and econometrics. And it's almost just playing to the strength of the rise of data volume, as well as the rise of computes capacity. So you'll see all the big players releasing their own versions of GPU. Where as I said, ML is all about numerical transformations, where things are just vectorized or described as matrices, where a couple of years ago-- and that's why Amphibia and all those players are doing so well nowadays. Is that we discovered that just optimizing for this graphical kind of like video game and graphical design use-cases, all those operations are on the scale of XY of your high and growing resolution of screens. 

RAF 35:13 

These are the exact same transformations you would want your deep neural nets to perform and [inaudible] over to refine the accuracy of those algorithms. So NLP is a special, very special subset. Because this really links to us having conversation. It's almost tackling the ultimate, kind of like human-like, kind of like reserved space of knowledge and discovering the intent and the sentiment, which is just this kind of like very noble and high reference point. Where in fact, the actual accuracy and the sophistication of the NLP advancements that are out there with some great Python and RLI [inaudible] can easily be embedded within Alteryx pipelines. They are just very accurate. And the great surprise on our side is how accurate these are. And they not only automate the task of looking and almost providing these conversations with our customers, where they tell us that something is wrong. And then we can respond almost instantaneously. And not only this, we can probably take this further and preemptively forecast some patterns to notify our contact centers to learn about certain patterns just growing in the background. So it's a really exciting space, really exciting. 

BRIAN 37:07 

And there's one more that you mentioned, that I think we should talk about is neural networks, right? And my understanding of a-- so I think when people hear the phrase neural network, they think Terminator, right? They think, "Oh. It's a human-like artificial intelligence or a machine learning." And I think actually-- again, correct me if I'm wrong. I'm totally just playing the layman here. And I'm sure people are going to listen to this and tell me I'm wrong. And that's totally cool, right? But my understanding of a neural network is that most machine learning models or artificial intelligence think like a computer, right? It's using the 1s and 0's. It's doing what it's designed to do, in terms of taking structured and unstructured data, doing some compute on it, and providing an output, right? And what a neural network is, is kind of the-- not the reverse, but it's training a machine to attempt to think like a human brain does, which is potentially less rational, right, in terms of just strict 1s and 0's. There's more variability in it. Is that a correct description? Or am I wrong there? 

RAF 38:18 

No. That's absolutely fine. There's probably still a lot more to describe the space. The top-level analogy is that neural nets really do represent the neurons in the brain. And the basic fundamentals of how our brains work with the signals that propagate and backpropagate to refine higher-level strength of signals to provide some action, let's say. So neural nets are really great at tackling all the abstract data, where there is high level of ambiguity. Let's say image processing and categorizing images as cat, dog, hot dog, whatever else, discovering the different parts, discovering the location of text on the image. Again, with the scale of things, so pixels, the ambiguity, and the kind of like abstractual elements of-- historically, photography is probably just a couple of hundred years old, I would imagine. It's almost like a made-up thing. And it's not naturally occurring in nature. 

RAF 40:01 

So it's almost going after those abstract solutions that are highly efficient at just discovering those hidden patterns behind the complexity of the item entity on hand. Again, there is probably more and more theory. But specifically, deep neural nets, with the use of GPU acceleration or Google's TPU and multiple other flavors of those, are arriving at just surprising level of accuracy and insight, to the point that some of the natural language processing scenarios involve making language mistakes on purpose, just to imitate the human element. Where in fact, the accuracy is so high, it's almost beyond human level, with certain applications. So it's a really exciting space. 

BRIAN 41:14 

Yeah. Very cool. And what we'll do, is we'll drop a link in the show notes to a friend of the show. Jason Mack published a blog post a couple of weeks ago about this very topic and understanding what is AI, what is ML, what are all these things, so kind of further reading for anybody kind of interested in these topics. I wanted to also talk about your Agile manifesto. So we'll put a link to this, as well, in the show notes at community.alteryx.com/podcast for our listeners. Tell me about your manifesto. What is this? And why does it mean so much to you? 

RAF 41:53 

Absolutely. Just to make things straight, it's not my manifesto. It's the reference point-- so we all live and breathe the agile ways of working and principles that are pronounced with the Kanban boards that we populate with the backlog of workloads. The daily scrum kind of like setups for how teams just organize themselves and operate. What I discovered is that, when I started the actual behind the scenes foundation of this term, is that it's quite surprising that it already dates back to 2001, so it's nearly 20 years old. It's that a group of software guys from all different organizations got together and just brainstormed this, "Let's create this kind of reference point of how do we organize software development." And there was values and principles that they formulated and agreed upon, that they actually formed the foundations of these tools that we operate on a daily basis. And they're very strangely insightful and almost universal. So I'd like to probably just quote 2 of the 4 key values, and one of the 12 principles, and explain how it actually aligns with Alteryx. 

RAF 43:47 

So first of all, it's individuals and interactions over processes and tools. And this may sound contradictory because Alteryx is a tool. So what does it mean? And I have an experience from not long ago, probably just a couple of days ago, where I said, "Okay. So I've never used Spatial. And I just passed by my core certification. Let me do my homework, and just get my mindset into Spatial." So what I did, I actually reached out to the very team there that's mastering this toolset space. These guys look after market development. So they define where we create new stores, and how do you kind of like optimize the location element. And I was very surprised, in this very meeting, where we almost came into the meeting room as strangers-- and this probably links back to this reference I made, "Excuse me, do you speak Alteryx?" And the answer was-- I mean, I knew what the answer was. But this reference point allowed us to connect instantaneously. We got the full access to all the databases, provided all the security clearance, of course, and doing the right thing. But these guys just offered the whole spectrum of tools and knowledge and experiences through Alteryx workflows, through galleria apps, to some macros with some really pretty nice icons. And I was just like, "Wow. They really live and breathe this Agile principle." And Alteryx is the foundation to all of this. 

RAF 46:00 

And maybe let me reference one more. Responding to change over following a plan. Again, just playing an Alteryx pipeline, where you plan to cross 5, 10 different sources, you're provided with this fine level of detail of following every single record. And the thing is unbelievable. It's actually the mindset that I'm trying to bring into the big data world, where you may not care so much about the precise record that count. You almost operate at the high level of abstraction where same things are just approximized. So this principle with Alteryx allowing to almost just welcome new requirements as we go and play, and you surround your data pipeline with a galleria app, and you blend Tableau-- and probably the last point, we didn't cover this. I like to think of the data ecosystem as this blend of three key components, where Alteryx is covering the data wrangling connectivity side of things. That the second aspect is covered by tools like Tableau or Power BI, where it's still under a value to visual inside discovery, and how big of an impact it can make with this high velocity, variety, and volume. And this element of the high pace of data flowing, and IDS flowing. 

RAF 47:58 

And the third element is not really discussed-- that's maybe a problem. So I want to leave you with maybe just a little problem to take forward, maybe in some future podcast. So data storage, but it's not the original data that you connect to as the data source that resides all over the place. It's the data storage that you create as an outcome of your processing that can link back to those inputs data source. And you can be doing some sandbox storage or develop a discrete space for the output of your model. But I think that there is a great need to tackle the third element that's being self-sufficient and empowered with the connectivity and data-wrangling capabilities, that the actual storage doesn't really need to optimize so much for this normalization of data, where things are broken apart. In fact, Tableau likes to play against self-contained data blocks. And the power is shifting towards the end-user. That also means that the actual capacity and capability of designing the storage facilities should follow along. But it's not the case. It's one aspect that we continuously struggle. Almost describing the need for a robust data storage as a result of having all those magical powers with Alteryx and Tableau. 

BRIAN 49:57 

Yeah. So that's amazing. We will link this manifesto in the show notes, for anybody who wants to go read it. I recommend it. I think there's a lot of really interesting practical tips in there that, whether you write code or do any work in data or not, there's just some really wonderful kind of insights in this that really can help you think about the way you work and the way you approach things. And I'm sorry that I, at the top of this segment, attributed it to you. But just to make that right, while you were talking, I went to Wikipedia and edited it, and attributed it to you. So you are now the father of the Agile manifesto. I hope that's cool with you. 

RAF 50:36 

Oh. Wow. Thank you. That's amazing. 

BRIAN 50:40 

Awesome. All right. So let's segue to our community picks. What's been great? What should people go take a look at out there on the world wide web, as it were? 

RAF 50:52 

Absolutely. I think I've got two or three podcasts, as well as one book title. So the podcasts that fill my daily commute are Techmeme Ride Home. Another one is Data Skeptic. And the final one will be TWiML&AI, which stands for This Week in Machine Learning and AI, which I'm trying to [inaudible]. So these are the great kind of like reference points that inspire me every. And I come to work charged with all those ideas. And I load Alteryx and just try things. The book title that I'd like to reference is Kevin Mitnick's, Ghost in the Wires. And it may be slightly controversial. It's tacking social engineering, which I like to rephrase as social engagement. And from the time perspective, I actually consumed this book two or three years ago as an Audible on my daily commute. Is that I liked the reference. So Kevin is the world's most wanted hacker, or used to be a couple of years ago. And now, I like to abstract and to take a step back and think as-- he was a very curious man. He was someone that tried all these new ideas. He was tapping into telephony and the kind of like early days of internet. But behind all of those great, great stories, you can almost just take the learnings of being brave, being curious, trying things, and owning your career pathway, where Alteryx is such a great companion to all of this. So yeah. So these are my community picks. 

BRIAN 53:08 

Great. Tons of good stuff for us to go listen to. Mine is, so a couple of weeks back, I was scrolling through this-- it's this small, this kind of small community that I found that I don't know if many people know about it. It's called Twitter.com. But anyway, I was scrolling around through this community, and I stumbled across this tweet from vox.com. And we'll put the link to this in the show notes. And they had this really cool video. It's like three or four minutes long. And they're explaining how here in the states, of course, we have the NFL, how do they put the line on the field, right, digitally. And they also discussed-- many people don't know about this, or don't remember it. But back in the early 2000s, I think it was, the NHL, the National Hockey League here also had a similar project, where they were-- they called it glow puck. And they were trying to make it kind of like the video games. So when a player shot the puck really hard, the puck would have this tail. And it would glow red, so you could see it better on TV, was the idea. Anyway, the video kind of describes the technology and really the data behind all of this, and how they have sensors around the arenas that take in the location of either the puck or the players or the field itself, or whatever the case might be. And how they then, take all of that data, and digitally impose the imagery on the screen in real time. And I think that's the key to kind of the-- the interesting part about it, everybody knows that movies and television shows and things that we consume today are heavily CGI. But think about doing that actually in real time, and having that render on the fly during the live broadcast, right, so really interesting kind of data aspect there. We'll put the link for folks to go check that out. 

RAF 54:57 

Amazing. 

BRIAN 54:59 

Yeah. All right. Raf, this has been great. Thank you for being on the show. I really enjoyed the insights. We went a little bit long. But I think tons of great content for folks there. So thanks so much. 

RAF 55:11 

Thank you, a great pleasure. Thank you for all the great effort of putting this and keeping the dialogue of this high quality across all your editions of the podcast. Thank you so much. 

BRIAN 55:26 

[music] Yeah. Happy to do it. We'll see you soon. [music] Thanks for listening to Alter Everything. Go to community.alteryx.com/podcast for show notes, information about our guests, episodes and more. If you've got feedback, tweet us using the #altereverything or drop us an email at podcast at Alteryx.com. Catch you next time. [music] 

BRIAN 56:08 

So Raf, you've listened to the podcast quite a bit. You've probably heard me waxing poetic with other guests about space travel and aviation and other related things. I heard you earlier, I think right at the top of the show, you were introducing yourself, and you described your name and how it's spelled. And then you said, it's something like Royal Air Force. And I knew, in that moment, that we were going to bond in the after-show over aviation. So talk to me a little bit about aviation. Tell me about how you got into that, and describe your passion for me. 

RAF 56:47 

Absolutely. So I was quite fortunate to be part of this great startup that did air cargo for a subcontracting for some of the big brands, like DHL, UPS. And I looked after the IT space for this enterprise for a couple of years. But the best times were when I was taking part in some of the technical evaluation and testing. So there are two stories that were really just amazing, and the memories of which stay with me until today. So the first one was where we had the chief technical guy ask me to go into the cockpit. So the aircraft was an An-26. It's a Ukrainian based cargo aircraft. So it's really [inaudible]. It's a propeller. So you get the experience of the sound and the true nature of the physics of flying. So we were testing the new propellers, and I was asked to hold down those two [inaudible] where they guy went full throttle almost the airborne speed. It could feel like when you forget your hand brake in the car, if you're still operating these nowadays. You feel the power kind of like gaining momentum, where there's this thin line that's almost holding this in the spot. And I was sweating the whole thing. So when you [inaudible], it all went very well. But with all these great colleagues out there, they explained all the physics, how if I let it go, what would happen. I'm not going to describe this, but it's almost like physics of a bullet. So all was in my hand sweating, in my legs, and I was just petrified on the spot. And I was so glad at the end. 

RAF 59:21 

The second one that I will share with you. So again, we were testing some routine of some service maintenance procedure. And we were flying at very low altitudes. I wasn't introduced into the script of the procedure, but it went like this. So the first engine turned off. We tilted to the side. I'm like, "Whoa." Like, "What's going on?" And I [inaudible] that we'll not only be turning practically this one engine, but will turn the only other remaining. So after a brief while, we were just gliding this massive metal construction just purely on the rules of the physics of the gliding pathway. We went back home, everything was fine. I was glad, again-- so scared inside of myself. I'm like, "Okay. So we are approaching the landing." As I would imagine the final step, you then just go down with your speeds and off you go. But what I didn't know at this point is that they're actually doing the touchdown procedure. That's when you essentially touch the ground with your wheels, but then go back up again. And this was, again, something just brand new experience and so exciting just looking back, but so scary at the same time. Those are the kind of like all those cherries on top of the delicious dessert I had just working with the aviation kind of like guys. 

BRIAN 01:01:25 

That sounds amazing. So I'll share a couple of stories. So I grew up going to air shows. I was born and raised in Indiana. And not too far from a place called Battle Creek, Michigan. So for some of our listeners out there, they may know where Battle Creek, Michigan is. Of course, it is the home of our friends at Kellogg's. And so, every year, in Battle Creek, Michigan, they have one of the largest air shows in the world. I think it's the largest in the United States. I'm not sure if it's in the world or not. But it is a massive, multi-day air show, where hundreds of planes fly during this time and do maneuvers. And of course, you have the Thunderbirds and the Blue Angels and these air demonstration groups. So I grew up doing that. And here's a pro tip for everybody. If you ever get a chance to go to an air show, you should definitely do it. But here's the tip. If you can find someone where you can rent a scanner, one of the coolest things about going to an air show, if you're able to rent a scanner, is you can plug in the radio frequencies that the pilots use, specifically if there's a demonstration team like the Thunderbirds or the Blue Angels or the Snowbirds that are from Canada, those kinds of things. And you can actually hear them communicate with each other as they fly. And it is one of the most incredible things that you'll ever hear. Because these guys are going hundreds of miles an hour, and they're literally inches away from each other. And to hear the precision with which they operate is one of the most amazing things. There's also videos of this on YouTube. So I guess, we can just drop one of those into the show notes. And you can hear their speaking cadence, and how they have this down to a science. So that's one story. 

BRIAN 01:03:12 

The second story and I don't think I've told this on the podcast before. But I also used to be into racing. And so one year, being an Indiana native, we would go the Indianapolis 500. And one year we went, and in typical Indiana fashion-- sorry Hoosiers. It was raining. And so the rain is pouring down. The race is clearly not happening. But, of course, one of the things of pomp and circumstance of auto racing is the flyover that takes place during the national anthem, right? And that particular race, a B2 bomber was the scheduled flyover. So this is the-- for those of you who don't know that are listening, this is the Flying Wing, the Black Wing. You may have seen it at a bunch of movies, like Broken Arrow, right? It's the stealth bomber. And so this thing was on its way. And what people don't know about that is, those only fly out of, I believe-- I don't want to get this wrong. I'm pretty sure it's Alabama, is where those are stationed. And it's the only place they fly out of. So when they get scheduled to do a flyover for a sporting event, whether it's a Super Bowl or a racing event or something like this, they fly all the way from-- again, I believe it's Alabama. It's a southern state in the United States. So that's a long haul for them to fly all the way from there to, say, Indianapolis, Indiana, right? So they have to set out on that journey long before the race is on, or the weather does what it's going to do. 

BRIAN 01:04:42 

So the long story short is, it was raining. The race was clearly not going to happen. And they had delayed it. And so we were walking back to our car to just kind of get some refuge from the rain. And we were underneath this overhang, kind of in a parking structure. And it was right around the time when the national anthem would have been taking place, had the race not been rained out. And all of a sudden, we hear-- and the clouds, by the way, the clouds were probably just a couple of hundred feet off the ground, I mean, very low ceiling. And all of a sudden, my dad and I are walking under this overhang. And we hear this low rumble. And it was like nothing I'd ever heard before, just this very low, bassy kind of rumble. And it's just pouring rain, and the cloud ceiling is low. And we hear this rumble. And we step out from underneath this overhang. And right as we step out, the B2 bomber swoops down below the clouds, and it's like-- I mean, it had to be 500 feet off the ground or lower. This thing was low. And it's just cruising right below the cloud layer. And it kind of went over where the track would be. And then he just cut right back up into the clouds. And it was just one of the coolest things I have ever seen in my entire life. And I was always interested in that stuff. But that moment cemented my kind of wonder and love of aviation. And long story short, I was planning on kind of being an aviator at some point. But I'm a tall guy, and so that didn't work out. But anyway, really interesting stuff and really cool to hear your love for aviation, as well. 

RAF 01:06:25 

Absolutely. No. Just amazing stories from you, Brian. I was quite fortunate to have witnessed Antonov An-225 Mriya, which is Ukrainian for dream, that actually probably can take some sort of Boeing constructions inside, and it's meant for just moving tanks or some other heavy machinery. And this thing it's got probably hundreds wheels, six engines. It's just the biggest thing ever. So I've got the same passion for this sort of thing. So I think, as you said, we are bonded on this one. 

BRIAN 01:07:19 

So one of the things I've always wanted to do-- it's really expensive. I'll just say that off the top. But NASA has this thing called the Vomit Comet, which is how they train astronauts on weightlessness. And it's basically a Boeing 767 or something like that. It's one of the big airliners. But it's stripped down, and the inside of it is all just padded. And they basically go straight up with it, and then they go straight down. And you get several minutes of weightlessness, right? And so they've been doing that for years. But there's now a private company here in the United States. I don't remember what it's called offhand. We'll have to find this and put the link in the show notes. I think it's called Zero G. But anyway, they actually have one of these planes that they've outfitted. And they fly around the country. And you can pay, as a citizen, to go do this. And it's very expensive. So I'm waiting to hit the lottery before I go do this. But here's my commitment to you. Some day, if one of us hits the lottery, or something like that, you and I should definitely get together and go in the Vomit Comet and experience some weightlessness. I think that would be just the capper to this whole thing. 

RAF 01:08:27 

Absolutely. That's a deal, Brian. 

BRIAN 01:08:30 

Deal. Done. 

RAF 01:08:32 

Done. 

 

This episode of Alter Everything was produced by Maddie Johannsen (@MaddieJ).