Data Science Mixer

MaddieJ · ‎01-03-2022

As one of the first data scientists at Peloton, Ibby Syed has a passion for enabling others to access powerful data science tools and techniques in order to drive insights and learning.

Panelists

Ibby Syed - LinkedIn, Twitter
Susan Currie Sivek - @SusanCS, LinkedIn, Twitter

Topics

Cocktail Conversation

Ibby CC.png

If you had the opportunity to guest lecture a group of budding data scientists on just one topic - perhaps something you think everyone getting into the field should know and understand - what would that topic be? Why do you think that's so important?

Join the conversation by commenting below!

Transcript

Episode Transcription

IBBY 00:00	I think for anyone who's listening, I'd love to pub that if you're interested in chatting-- one of the things that I love the most is helping folks who are building new data products, who are thinking about the data paradigm in a completely different way. So I mean, honestly, an open shout to all of your listeners is if you're building something interesting, if there's something that we can chat about where you have an idea, we love meeting entrepreneurs that are building interesting things, so we can onboard their products at Peloton. I love just geeking out over different ideas. And so if you're out there and listening, and you're like, "Oh, I'm working on something," or, "I think that this paradigm that we're using the data world is wrong and that we should change it to be something else," those types of people are always the most interesting to talk to, so feel free to hit me up.
SUSAN 00:42	Welcome to Data Science Mixer, a podcast featuring top experts in lively and informative conversations that will change the way you do data science. I'm Susan Currie Sivek, senior data science journalist for the Alteryx community. You just heard from Ibby Syed who's the lead data scientist at Peloton and our guest today. Ibby has had a fast-paced career in data science, especially in recent months with Peloton's rapid global growth and the demands that's placed on their data science teams. Ibby's experiences are impressive and exciting. And it's awesome to hear about all the ways he's made data science and analytics more accessible to others at Peloton and beyond. As you can tell from what you just heard, he's generous with his knowledge and time and truly passionate about the field. Let's get right into this great conversation. Well, Ibby, thank you so much for joining us today on Data Science Mixer. I'm really excited to have you here and hear about all the work that you've done in data science and now at Peloton. So I would love to hear a little bit about how you got into data science and how you ended up arriving at Peloton doing the work you do today.
IBBY 01:47	Sure. It's a bit of a windy story. I guess I joined Peloton pretty early, from a data science perspective, as one of the first data people there, but I got started with my journey probably in college. So I was a junior in college, and a couple of friends-- a couple of friends of mine had started a company, and I ended up joining as a late cofounder. We were experimenting with sort of a couple of products that we were building, but the way that we made money was by acting sort of like a software consultancy. We take companies or individuals that wanted work done, do some of that design work with them, and then manage an engineering team to do that, so effectively, very close to a design firm of a development shop or-- some of them were data-related. Some of them weren't data-related. But as you can tell from my resume, I definitely don't work there anymore. I'm not a part of a startup. And so that startup didn't actually end up passing. We ended up closing shop midway through about a year, a year and a bit later, very unfortunate. But one of the projects that I was working on at the time was data science-related, specifically NLP space. And if you ever get a chance to meet him-- David Cancel, he's the CEO of Drift out in Boston. I got a chance to talk to him, and he was really, really interested in that project. He was like, "Hey, why don't you come on board to Drift and help us sort of develop this?" And so my senior year, that company had shuttered. We had a little bit debt we had to pay off. And so David Cancel, thanks to him, he gave me the chance to sort of build out some of the data stack and some of the NLP stack over at Drift.
IBBY 03:17	And I worked with a sub-team that they had there that was out of the MIT Innovation Lab. They're working on some sort of AI. Got started there. That was where I got shucked to Graham Stanton who was one of the founders of Peloton. And they were starting a data science team. And sort of I met him, and we've talked about what we were working on at Drift, what some of the interesting problems were at Peloton. And it was like, "Hey, why do you come in for an interview?" I flew down-- or trained down from Boston to New York, met some of the team there. And they were working on a lot of really, really interesting things around using data science to promote acquisition, to promote retention, making sure that onboarding, how we onboard the customers, was as optimized as possible; they had a great customer experience, trying to develop a lot of products internally around using data to make decisions. And yeah, I joined two, two and a half years ago and got to build out a lot of the first data stack, SQL transformations, the BI layer, got to build out machine learning capabilities on the operation side, and gotten to build out sort of AutoML and research capabilities, which has been super, super fun. Absolutely love Peloton.
SUSAN 04:26	Awesome. That sounds great. I love how every journey that we hear from our guests is always a little bit windy in some interesting and fun way. So you're in good company there. So you mentioned that you were one of the first people coming on board to be more dedicated to data science, specifically at Peloton. What was that experience like? And maybe out of that, do you have any advice for people who end up in a similar situation, potentially?
IBBY 04:49	Yep. Well, so hindsight is obviously always 20/20. But I should say I think the biggest advice slash-- the biggest learning that I got is if you're interested in data-- and this is just data. This is not sort of-- I don't think this can be expanded to tech as a whole. But if you're interested in data, being a part of building or rebuilding a young company's data stack is, in my opinion, one of the most important things that you can do. In terms of just the sorts of tech that we had to work with involves sort of when we were going public, the KPI generation and how we were going to develop levers around how KPIs were moving, building ETL and streaming pipelines to ingest data from different services, writing sort of scheme architecture and designing how data was going to look for the end consumer, i.e., internal data analysts, internal data scientists - what else? - creating a flexible sort of BI layer. We use Looker for people to have easy access to data. And then sort of building a machine learning stack for operations on top of sort of the data warehouse was a super, super fun part. We're working on AutoML, making sure that you sort of democratize access to data for everyone at the company, regardless of the languages that they write in, regardless of sort of how familiar they are with data. The best data product, in my opinion, is when you're able to ask a question, have sort of a decent understanding of how data should look, and then you can go to a platform that will, no matter what your background is in data, help you answer that question or if it's a really hard question, help connect you to somebody who can. And that's sort of our sort of data mission at Peloton.
SUSAN 06:20	Nice. And so it sounds like a whole lot for somebody who wasn't new to data science but new to the company to be taking on to do all of the tasks you've just described. Was that overwhelming at times? Was it just intellectually exciting and professionally exciting?
IBBY 06:37	I think the line between professionally exciting and overwhelming is who the company is surrounded with, right? So I think it's entirely dependent on the type of people that you work with. And the best part about Peloton is just how supportive the culture is there. I think it's very much we try to have a baseline of hiring really, really smart and talented people and then instilling a lot of trust in them. I think the best part about-- again, going back to my previous example, my advice for young people that are looking to get into this space is join a company that doesn't necessarily have a training regiment, right; you're not going through sort of three weeks or four weeks or five weeks of data-science onboarding, but rather, you have a great manager, a great sort of set of people that you work with that are willing to sort of give you the keys to a part of the business that's really important and say, "Hey, we have a lot of stuff going on. Here's something that you can work on. Here's something that you can really improve yourself in." And getting that trust handed to you, it was really beneficial for me. And I think it makes it so you're at the very beginning of your career set up for success, and you're set up to build something really, really amazing. I think if I look at where our stack was two and a half-- two and a half, three years ago and where it is today, we've gotten the chance to build so much.
SUSAN 07:55	Yeah, yeah, absolutely. Yeah, it'd be really exciting to see that entire evolution from the ground up. So Peloton, of course, folks know the name. It's a household name at this point. So over the last year, there's been massive growth, especially during the pandemic and with a lot of global growth. So what has that experience been like as you've been working on data science projects at the company?
IBBY 08:16	Yeah, the biggest one was just the rapid increase of questions that we got, right? So when the pandemic hit, our sort of entire worldview changed. Demand was going through the roof. And I mean, we still have really, really high demand, but back then, we just weren't set up to scale nearly as quickly. And so it was a lot of questions on, "All right, so have we expanded our total addressable market?" right, "Who else is buying Peloton maybe that wasn't buying it before? Should we be making considerations on safety? Should we be making considerations on--" all of these sorts of really disparate questions that arise when your business sort of takes a turn overnight and becomes this-- it was always popular but something just even bigger than anything that we could have imagined. Nobody predicted the pandemic, least of all, us. And some of the tailwinds that we got from it were-- there was a lot of interesting questions that came up. So the biggest thing that we had to do was we had to scale our team incredibly quickly. So at the beginning of the pandemic, right, it was right after where we were about 5. And we're almost at 50 now. So it's--
SUSAN 09:21	Wow. It's amazing.
IBBY 09:23	Yeah, huge. There's a lot of questions around supply chain optimization, understanding demographic shifts. And we had just launched-- midway through the pandemic, we launched our new product, our Bike+, and trying to figure out sort of, "Hey, where do we put this-- where do we place in the market? What type of customer is really going to love the Bike+ over our sort of standard bike offering? Is it people who are already part of the Peloton ecosystem? Is it people who are interested in cycling more than the customer that we had before?" Those sorts of questions, I think, they came up a lot around when we launched the Bike+. We wanted to make sure that it's the [resounding?] success that it is. So yeah, hiring is definitely the biggest one. We grew every aspect of our business, and so all of the stakeholders that we have-- the way that we've set up our data team is it's relatively centralized with-- there's a lot of analysts and other data scientists that sit on other teams, but we're, by far and away, the largest data org. And the number of stakeholders that we had just grew by about three or four X overnight, and so we had to scale our team to make sure that all of the questions that everyone was coming to us, all the new questions that we were staffed, to be able to handle those.
SUSAN 10:34	Yeah, definitely a challenge to get enough people on board to take care of all of that, for sure. Interesting. So now, I have to ask since you just mentioned the products, are you a cyclist, a runner? Are you--?
IBBY 10:44	I'm a cyclist. Yeah, I am. I definitely should do it more than I do.
SUSAN 10:52	I think we all have that with working out.
IBBY 10:54	Yeah, I think with working out. I try to work out a couple of times a week. It's almost always an Alex Toussaint ride or a Emma. I love trying new instructors, especially as we've brought on a lot of new instructors over the course of the last couple of months. It's always super fun. But yeah, I am not a runner. I wish I was a runner. Unfortunately, I don't know. I'd never got into it. But I love the bike. I love the bike. I got one for my family. I have one myself. And it's a really good way of-- a lot of my friends have it now, too, and so competing with each other, making sure that if we have a ride that we absolutely love, we'll have a text chain where it's like, "Hey, this is a ride that we really--" Our teams, too, actually. Internally, at Peloton, we have a team ride that we do every Friday, which is super fun. My coworkers are just so much more in shape than I am. It's hilarious. It's just so hilarious. Oh, you know what, I'll be like, "Oh--" I'm like, "I'm doing well. I'm doing well for myself. I'm doing my own PR." And then it's just like I'm the bottom of the pack.
SUSAN 11:52	Oh, that's so funny.
IBBY 11:54	It's really funny.
SUSAN 11:56	That's cool. I love the idea that all of your friends' data and your family's data and your team's data and everything, it's all wrapped up in your work too. It's kind of cool in itself. So that's a lot of fun.
IBBY 12:06	Yeah, yeah, we definitely, definitely privatize that sort of stuff. So I can't tell who's who. But yeah.
SUSAN 12:10	Oh, for sure. Yeah, yeah, awesome. So I'd love to hear maybe about one or two data science projects that you've done at Peloton that you're maybe especially excited about or proud of that you can share publicly?
IBBY 12:23	Yeah, sure. I think the two most interesting ones would be-- the first would be furthering sort of our AutoML capabilities. So going back to what I said before, we want to really make sure that we democratize access to data and democratize building projects on top of data. And one of those things we believe is sort of operations machine learning and operations data science. So making sure that somebody who has a really, really good understanding of the business can do, quote-unquote, "feature engineering" without necessarily needing to have a Python stack. There's actually a lot of companies that are coming out these days that are doing this. [inaudible] is one that is really, really cool. But their entire mission, and this is something that we've built internally, is to take people who are very good at SQL, very good at structuring datasets together-- it can be, honestly, somebody who just uses Looker, Alteryx, or Excel, or a tool that can sort of aggregate data into a dataset. People who understand the business and the data for the business really, really well can do effectively the feature engineering part of data science and then quickly build propensity models, quickly build classification models, quickly build some of the-- even with time-series data, quickly build sort of a regression model. That's, I think, one of the most important projects that has found success for us where we've gotten the chance to-- a lot of folks at the company have stepped up and started using stuff like that, whether that be AutoML-- the other one is sort of a pseudo-related example, but it's auto research effectively, so using the same principles to understand the changes between different cohorts. So an example of this would be sort of templatized notebooks or templates, basically, where folks can come in and-- let's say that they're looking at two user groups, right? They're like, "All right. Here's a user that bought a Peloton bike a couple of years ago. Here's a user that about a Peloton bike more recently. How is the usage pattern different? What kinds of classes are they enjoying more?"
IBBY 14:20	Yeah, so we've gone in, and we've created sort of this methodology to use automatic research-- to do automatic research where, let's say, you're trying to look at the differences between two cohorts, right, one that bought a Peloton or a user that had their Peloton journey start quite early, whether that be 2016, 2017, or someone who had started a little bit later, somebody who joined during the pandemic, "What sort of classes are they doing?" right, "Are they taking advantage of some of our newer offerings? Do they only take cycling classes? Are they taking meditation classes?" It really helps us understand, "All right. Where should we be developing new class types? What are people enjoying? How do we sort of predict what type of content folks are going to be relating to?" and basically give folks at the company a way to define two cohorts or define a test or do an observational study with internal data and understand very, very quickly how two groups of users-- or a study differs. So that's been another sort of infrastructure thing that we've implemented that has been very, very, very interesting. It just takes data from all of the different disparate data sources that we have, collects them all into one location, and then does a sort of visualization-- has a visualization element on top of it that other-- we'll do a bar chart or histogram or sort of a network analysis on different types of users. So it gives us a good way of figuring out where to sort of build next.
SUSAN 15:46	Awesome. Yeah, yeah, definitely. No, that sounds very cool. And I'm curious, with both the AutoML and automatic research projects, when you've got people across the organization potentially using those, how do you get people to buy into those, to learn how to use them, to feel comfortable with them? How has that process worked for you?
IBBY 16:08	Yeah, training is absolutely one of the most important things that you can do, right, trying to make sure that you set up technologies that folks will inherently use but also that there's a very, very limited amount of switching cost from the way that they're currently working on it to using the new system. And I think the best way is-- we just have a library of internal videos that we use to describe and show off what we're building. We have an internal channel where when somebody is building something new, they post about it; they say, "Hey, this is what I've built. This is sort of where it fits into the larger Peloton ecosystem," whether that be internal or external, "And here's sort of a brief video going over what it does." And so we do that in a little bit of a longer fashion. A lot of our sort of onboarding for new employees has a lot of these sorts of videos for tooling and a lot of documentation to where when somebody comes in, they can quickly go, "I'm trying to use Alteryx," or, "I'm trying to query something from Redshift. I'm trying to understand our data models." They sort of have a series of videos and documentation that they can use to get to that.
SUSAN 17:15	Nice. Pretty cool. It sounds like a lot of effort going into creating all of those materials but also a lot of potential payoff for having people have immediate access to that.
IBBY 17:22	Yeah, it makes it so you end up letting people answer questions very, very quickly and effectively, and there isn't necessarily a time sink from somebody else who's been at the company longer. It's perspective, right, so they're not-- obviously, we promote working together. We promote a lot of collaboration. But if it's a simple question, we want you to be able to answer it yourself. That's a really, really important part of getting things done quickly.
SUSAN 17:48	Yeah, absolutely. And especially with, as we were just talking about, the incredible growth and expansion that you were dealing with, too. So offloading some of that onto individuals would be definitely a help in that process. Cool. So I noticed also looking through your resume and records out there that you were also teaching data science until pretty recently at Columbia. So I'm just curious what that experience was like for you and if that informed your day-to-day work at all, had any other impacts on your data science professional endeavors.
IBBY 18:17	Yeah, yeah, I wasn't technically a Columbia employee or anything. But Columbia actually has a data science boot camp that they offer. They had to do with a partnership with the company. And I was a TA for a couple of cohorts, and then I got a privilege of a lot of teaching time last winter. And honestly, for folks who work in the data space teaching and for those who have been there longer than I have, you can probably get an adjunct role somewhere, teach different groups of students at different parts of their career, whatever. It's a great way to sharpen your skills and sort of help excite a new generation of data practitioners on sort of what's coming down the pipeline, right? If you're a part of the academic circles, or you're involved in boot camps, they're usually using very, very recent technology. A lot of folks either would read papers, or they'll understand, "Hey, this is a rising technology that's coming through the pipeline. We should develop our curriculum around it." And so one thing that was great was it allowed me to learn, honestly, a little bit more about the different types of data stacks that are at different types of organizations, right? So whether you're working sort of farther upstream; you're working in sort of the cache, the Redis world, whether you're working on the E and L part of ELT, rather than just the transformation part, and sort of helping students find what interests them within that sort of data stack world and saying, "Oh, this is the type of job that you might really love," whether that be machine learning engineer or data scientist, right, so people who are more on the analysis and model development side of things, or a data engineer or an analytics engineer or a data analyst or folks that want to end up doing product or visualization-related things, you do everything from Mongo and unstructured, no-SQL databases all the way to D3 and JavaScript implementations, which is really, really fun. It's a great way to sharpen your skills. It's a great way to excite people. It's a great way to honestly share your knowledge.
SUSAN 20:14	Nice. Yeah, that's great. And then certainly, I've experienced that as well that as soon as you can-- what's the saying? If you can teach it, you actually understand it, something like that. And so yeah, that's cool. So--
IBBY 20:24	Yeah, yeah, what's really fun, sorry, was to go back and understand a lot of the math around sort of deep learning, around some of the certain machine learning algorithms because you learn those in undergrad, right? I had a class where we just had to implement a lot of the very basic machine learning algorithms on paper. But then when I started my job and when I started working in it, a lot of it's been abstracted, and you sort of remember, "Oh, my goodness, this is exactly how this works. This is sort of the process--" There's a recursive algorithm somewhere in here that you just run over and over and over again until you get to the result. And it's really fun to realize that the next day, you have to teach a class on it, go back, read through the textbook that you haven't picked up since undergrad, and go, "Oh, yeah, I remember exactly how this works," and show folks that. There's obviously a lot of folks that work in data, right, whether that be on a research team at a big company or whether they're doing deep tech stuff at a very small company that they work with the fundamental unabstracted layer of it every single day. But a lot of times, you just sort of don't get that when you're doing the more engineering side of things.
SUSAN 21:34	Right. Right. Yeah, pretty cool. No, it sounds like a neat experience. So you mentioned that the boot camp was teaching some of the kind of cutting-edge stuff that's coming up right now in the field. What are some things that you're excited about in the future of data science, some things that really intrigue you that you would like to explore further?
IBBY 21:55	Yeah, I think the biggest benefits that we see and sort of my thesis on where this is going to go is we're going to end up abstracting, I think, a lot of the development and deployment aspects, so the more computer-science elements of data science, into disparate sort of tooling, right? So you see this with the rise of things like Kubeflow. There's a lot of open-source projects that are designed to sort of take the development that data scientists do and then either interpolate the code and create microservices out of it to where you can actually use it in development where-- I think Alteryx actually has a pretty cool set of tools where you can run workflows that are around data science or machine learning and sort of abstracting a lot of the work that data scientists do because a data scientist fundamentally just isn't an operational engineer a lot of the time where when you hire a data scientist, you want to make sure that you're hiring someone who understands feature engineering, who understands hyperparameter tuning. You want to find somebody who has a fundamental knowledge of all of the different types of algorithms and what to use for a specific type of question. What you don't necessarily always find is somebody who understands how to deploy that somewhere, right? And so there's obviously the rise of MLOps folks. But I think, especially for those younger teams, having tooling that will help you deploy and use some of that modeling work that you're doing for research purposes in production is going to be a really, really important thing.
SUSAN 23:24	Yeah, yeah, definitely. Cool. So I have one question that I always ask to our guests that we call the alternative hypothesis segment of the show. And the question is what is something that people often think is true about data science or about being a data scientist but that you in your experience have found to be incorrect?
IBBY 23:44	I think the biggest-- I think the biggest fundamental thing that most of my students actually, when I was teaching, learned is that after they graduated from the boot camp, and they later went on and started their careers in data science, or they took their careers in one part of their company and moved it over to being more data-driven is-- one of the most important things that an organization can do is develop a really, really, really solid BI layer that-- I think that most data scientists come in, and they say, "Oh, I'm going to be doing a ton of deep development," right? You're going to find a problem. You're going to develop a machine learning model for it. And you're going to, A, set it, and it's going to solve all of the problems you've ever seen. A, that's not really how it works. You have to go back a lot of the time. You have to make sure you're retraining your models; you're looking at a lot of the fundamental assumptions that you're building off of. And a lot of the work that you're going to end up doing, at least when you start, is going to be a lot of solving and answering really low-hanging-fruit problems. And I think the biggest thing that I hear from data scientists that have worked for a couple of years now is, "I really wish I could get away from having to write really, really simple queries. I really wish I could get away from doing some of the more sort of lower-hanging fruit sort of work, so I can focus on making a bigger impact."
IBBY 25:01	And the best way to do that, in my opinion, is to build a really, really solid BI foundation, making it so business users at your company can go in and say, "All right. I have the necessary tooling to be able to answer this question. I know where to go to answer this question. I know where the data lives. I'm able to sort of figure out all of the different types of scenarios that could sort of push this in one direction or another." And yeah, I think a solid BI layer and building automations is really, really important for that, right? Whoo-hoo, Alteryx. Just like the ATM, right, it revolutionized getting money and sort of left bankers' time to do more important things. I think that you can sort of view an automation and business intelligence layer as a way to sort of understand bigger-picture things around the company, especially when you're small, right, because a data scientist at a young company is going to be doing a lot of wearing every type of hat, right? That's something that you probably hear all the time where, "I'm a data scientist and we're small. I wear a lot of hats. I do everything from the E and L side of ETL all the way to deploying deep learning models in the cloud that we're going to use on our app side." And I think the biggest way to be able to fundamentally move forward as a data org is to make sure that the first thing that you're taking care of is having a really, really solid analytics layer to build off of.
SUSAN 26:20	What do you think are some of the barriers or challenges to making that happen?
IBBY 26:26	I think one of the biggest ones is just stack knowledge, right? I think a lot of what ends up happening is you join a job; you're focused on building out a very, very specific set of tooling, or you're focused on solving a very, very specific set of problems; and then you move somewhere else where you're doing a lot of the same sort of stuff. And I think parochialism is sort of the enemy of collaboration, right, where what ends up happening is we have either a lot of experts in a specific thing that don't necessarily always talk to each other, or we don't have someone who understands at least a small element of all of the different types of decisions that a business has to make to where they can connect all of the data together behind the scenes and make a really intuitive layer to help people understand that.
SUSAN 27:15	Yeah, great points. Is there anything that we haven't talked about yet that you would like to get in there?
IBBY 27:20	I think for anyone who's listening, I'd love to pub that if you're interested in chatting-- one of the things that I love the most is helping folks who are building new data products, who are thinking about the data paradigm in a completely different way. So I mean, honestly, an open shout to all of your listeners is if you're building something interesting, if there's something that we can chat about where you have an idea, we love meeting entrepreneurs that are building interesting things, so we can onboard their products at Peloton. I love just geeking out over different ideas. And so if you're out there and listening, and you're like, "Oh, I'm working on something," or, "I think that this paradigm that we're using the data world is wrong and that we should change it to be something else," those types of people are always the most interesting to talk to, so feel free to hit me up.
SUSAN 28:04	Awesome. Well, thank you for that. I know people appreciate that opportunity. Very cool. Well, Ibby, thank you so much for taking the time to talk with us today. I really appreciate it. And I know that folks are going to have a lot of interest in the things that you've talked about, and it will give them a little more insight into their Peloton experience, too, which is pretty cool.
IBBY 28:19	Yeah, yeah, thanks so much for having me on.
SUSAN 28:22	Thanks for listening to our Data Science Mixer chat with Ibby Syed. Join us on the Alteryx community for this week's cocktail conversation to share your thoughts. Thinking about Ibby's experience teaching, if you have the opportunity to guest lecture a group of budding data scientists on just one topic, something you think everyone getting into the field should know and understand, what would that topic be? Why do you think that's so important? Share your thoughts and ideas by leaving a comment directly on the episode page at community.alteryx.com/podcast or post on social media with the hashtag #datasciencemixer and tag Alteryx. Cheers. So one thing that I forgot because we were talking about the record button is I forgot at the very beginning to ask a very important question, which is that on Data Science Mixer, we typically have some sort of snack or drink or something with us while we're recording, so do you have anything special there with you today?
IBBY 29:23	I have a water, so I could stay--
SUSAN 29:27	That's a popular choice. Yeah, we got to be ready for all that cycling, so I assumed--
IBBY 29:32	Exactly, right? I feel like I've been trying to be more healthy, trying to load on sugary drinks. However, my usual go-tos are mango juice during the day or a yogurt drink. I just had a quinoa salad right before I joined-- right before I joined the call. That was sort of my lunch, which was really, really delicious. Yeah, no, today is just boring, no interesting drinks today. I knew that you were going to ask me this question, and I knew that I should bring something up. And then I went to my fridge, and there were two options. One was milk and the other one, water. And I was like--
SUSAN 30:06	It's all good.
IBBY 30:07	"--I'm just going to go with the simple water," because if somebody asks me what interesting thing I'm drinking, and I say milk, I'm going to get judged, I think. I'm not super worried about getting judged, but I am a little worried about getting judged.
SUSAN 30:20	Gotcha. No, no judgment here. And water makes a lot of sense. It's been a very popular choice lately, actually. People are apparently actually trying to work during the workday, so we do these podcast recordings, and they're not imbibing, I mean, really.
IBBY 30:33	Ugh, what? How? Why? Why am I not imbibing? It's a Friday. Why am I not imbibing? No, you're right. You're right.
SUSAN 30:38	That's true. That's true. And it's a little later over there. No, I just have some good old black coffee here, some nice French roast, so, much needed getting through--
IBBY 30:46	Oh, that sounds good.
SUSAN 30:47	Yeah, yeah, good stuff.
IBBY 30:49	Do you grind your coffee yourself and do the entire process, or?
SUSAN 30:54	No, as somebody who lives in Oregon, I have a really unsatisfactory answer to that question, which is I buy pre-ground French roast at the grocery store.
IBBY 31:02	I do too. I do too. Trader Joe's French roast is one of my favorite-- one of my favorite coffees, so yeah, very good.
SUSAN 31:07	Yeah, it's good stuff, for sure, for sure. Yep. Yay.

IBBY 00:00 I think for anyone who's listening, I'd love to pub that if you're interested in chatting-- one of the things that I love the most is helping folks who are building new data products, who are thinking about the data paradigm in a completely different way. So I mean, honestly, an open shout to all of your listeners is if you're building something interesting, if there's something that we can chat about where you have an idea, we love meeting entrepreneurs that are building interesting things, so we can onboard their products at Peloton. I love just geeking out over different ideas. And so if you're out there and listening, and you're like, "Oh, I'm working on something," or, "I think that this paradigm that we're using the data world is wrong and that we should change it to be something else," those types of people are always the most interesting to talk to, so feel free to hit me up. SUSAN 00:42 Welcome to Data Science Mixer, a podcast featuring top experts in lively and informative conversations that will change the way you do data science. I'm Susan Currie Sivek, senior data science journalist for the Alteryx community. You just heard from Ibby Syed who's the lead data scientist at Peloton and our guest today. Ibby has had a fast-paced career in data science, especially in recent months with Peloton's rapid global growth and the demands that's placed on their data science teams. Ibby's experiences are impressive and exciting. And it's awesome to hear about all the ways he's made data science and analytics more accessible to others at Peloton and beyond. As you can tell from what you just heard, he's generous with his knowledge and time and truly passionate about the field. Let's get right into this great conversation. Well, Ibby, thank you so much for joining us today on Data Science Mixer. I'm really excited to have you here and hear about all the work that you've done in data science and now at Peloton. So I would love to hear a little bit about how you got into data science and how you ended up arriving at Peloton doing the work you do today. IBBY 01:47 Sure. It's a bit of a windy story. I guess I joined Peloton pretty early, from a data science perspective, as one of the first data people there, but I got started with my journey probably in college. So I was a junior in college, and a couple of friends-- a couple of friends of mine had started a company, and I ended up joining as a late cofounder. We were experimenting with sort of a couple of products that we were building, but the way that we made money was by acting sort of like a software consultancy. We take companies or individuals that wanted work done, do some of that design work with them, and then manage an engineering team to do that, so effectively, very close to a design firm of a development shop or-- some of them were data-related. Some of them weren't data-related. But as you can tell from my resume, I definitely don't work there anymore. I'm not a part of a startup. And so that startup didn't actually end up passing. We ended up closing shop midway through about a year, a year and a bit later, very unfortunate. But one of the projects that I was working on at the time was data science-related, specifically NLP space. And if you ever get a chance to meet him-- David Cancel, he's the CEO of Drift out in Boston. I got a chance to talk to him, and he was really, really interested in that project. He was like, "Hey, why don't you come on board to Drift and help us sort of develop this?" And so my senior year, that company had shuttered. We had a little bit debt we had to pay off. And so David Cancel, thanks to him, he gave me the chance to sort of build out some of the data stack and some of the NLP stack over at Drift. IBBY 03:17 And I worked with a sub-team that they had there that was out of the MIT Innovation Lab. They're working on some sort of AI. Got started there. That was where I got shucked to Graham Stanton who was one of the founders of Peloton. And they were starting a data science team. And sort of I met him, and we've talked about what we were working on at Drift, what some of the interesting problems were at Peloton. And it was like, "Hey, why do you come in for an interview?" I flew down-- or trained down from Boston to New York, met some of the team there. And they were working on a lot of really, really interesting things around using data science to promote acquisition, to promote retention, making sure that onboarding, how we onboard the customers, was as optimized as possible; they had a great customer experience, trying to develop a lot of products internally around using data to make decisions. And yeah, I joined two, two and a half years ago and got to build out a lot of the first data stack, SQL transformations, the BI layer, got to build out machine learning capabilities on the operation side, and gotten to build out sort of AutoML and research capabilities, which has been super, super fun. Absolutely love Peloton. SUSAN 04:26 Awesome. That sounds great. I love how every journey that we hear from our guests is always a little bit windy in some interesting and fun way. So you're in good company there. So you mentioned that you were one of the first people coming on board to be more dedicated to data science, specifically at Peloton. What was that experience like? And maybe out of that, do you have any advice for people who end up in a similar situation, potentially? IBBY 04:49 Yep. Well, so hindsight is obviously always 20/20. But I should say I think the biggest advice slash-- the biggest learning that I got is if you're interested in data-- and this is just data. This is not sort of-- I don't think this can be expanded to tech as a whole. But if you're interested in data, being a part of building or rebuilding a young company's data stack is, in my opinion, one of the most important things that you can do. In terms of just the sorts of tech that we had to work with involves sort of when we were going public, the KPI generation and how we were going to develop levers around how KPIs were moving, building ETL and streaming pipelines to ingest data from different services, writing sort of scheme architecture and designing how data was going to look for the end consumer, i.e., internal data analysts, internal data scientists - what else? - creating a flexible sort of BI layer. We use Looker for people to have easy access to data. And then sort of building a machine learning stack for operations on top of sort of the data warehouse was a super, super fun part. We're working on AutoML, making sure that you sort of democratize access to data for everyone at the company, regardless of the languages that they write in, regardless of sort of how familiar they are with data. The best data product, in my opinion, is when you're able to ask a question, have sort of a decent understanding of how data should look, and then you can go to a platform that will, no matter what your background is in data, help you answer that question or if it's a really hard question, help connect you to somebody who can. And that's sort of our sort of data mission at Peloton. SUSAN 06:20 Nice. And so it sounds like a whole lot for somebody who wasn't new to data science but new to the company to be taking on to do all of the tasks you've just described. Was that overwhelming at times? Was it just intellectually exciting and professionally exciting? IBBY 06:37 I think the line between professionally exciting and overwhelming is who the company is surrounded with, right? So I think it's entirely dependent on the type of people that you work with. And the best part about Peloton is just how supportive the culture is there. I think it's very much we try to have a baseline of hiring really, really smart and talented people and then instilling a lot of trust in them. I think the best part about-- again, going back to my previous example, my advice for young people that are looking to get into this space is join a company that doesn't necessarily have a training regiment, right; you're not going through sort of three weeks or four weeks or five weeks of data-science onboarding, but rather, you have a great manager, a great sort of set of people that you work with that are willing to sort of give you the keys to a part of the business that's really important and say, "Hey, we have a lot of stuff going on. Here's something that you can work on. Here's something that you can really improve yourself in." And getting that trust handed to you, it was really beneficial for me. And I think it makes it so you're at the very beginning of your career set up for success, and you're set up to build something really, really amazing. I think if I look at where our stack was two and a half-- two and a half, three years ago and where it is today, we've gotten the chance to build so much. SUSAN 07:55 Yeah, yeah, absolutely. Yeah, it'd be really exciting to see that entire evolution from the ground up. So Peloton, of course, folks know the name. It's a household name at this point. So over the last year, there's been massive growth, especially during the pandemic and with a lot of global growth. So what has that experience been like as you've been working on data science projects at the company? IBBY 08:16 Yeah, the biggest one was just the rapid increase of questions that we got, right? So when the pandemic hit, our sort of entire worldview changed. Demand was going through the roof. And I mean, we still have really, really high demand, but back then, we just weren't set up to scale nearly as quickly. And so it was a lot of questions on, "All right, so have we expanded our total addressable market?" right, "Who else is buying Peloton maybe that wasn't buying it before? Should we be making considerations on safety? Should we be making considerations on--" all of these sorts of really disparate questions that arise when your business sort of takes a turn overnight and becomes this-- it was always popular but something just even bigger than anything that we could have imagined. Nobody predicted the pandemic, least of all, us. And some of the tailwinds that we got from it were-- there was a lot of interesting questions that came up. So the biggest thing that we had to do was we had to scale our team incredibly quickly. So at the beginning of the pandemic, right, it was right after where we were about 5. And we're almost at 50 now. So it's-- SUSAN 09:21 Wow. It's amazing. IBBY 09:23 Yeah, huge. There's a lot of questions around supply chain optimization, understanding demographic shifts. And we had just launched-- midway through the pandemic, we launched our new product, our Bike+, and trying to figure out sort of, "Hey, where do we put this-- where do we place in the market? What type of customer is really going to love the Bike+ over our sort of standard bike offering? Is it people who are already part of the Peloton ecosystem? Is it people who are interested in cycling more than the customer that we had before?" Those sorts of questions, I think, they came up a lot around when we launched the Bike+. We wanted to make sure that it's the [resounding?] success that it is. So yeah, hiring is definitely the biggest one. We grew every aspect of our business, and so all of the stakeholders that we have-- the way that we've set up our data team is it's relatively centralized with-- there's a lot of analysts and other data scientists that sit on other teams, but we're, by far and away, the largest data org. And the number of stakeholders that we had just grew by about three or four X overnight, and so we had to scale our team to make sure that all of the questions that everyone was coming to us, all the new questions that we were staffed, to be able to handle those. SUSAN 10:34 Yeah, definitely a challenge to get enough people on board to take care of all of that, for sure. Interesting. So now, I have to ask since you just mentioned the products, are you a cyclist, a runner? Are you--? IBBY 10:44 I'm a cyclist. Yeah, I am. I definitely should do it more than I do. SUSAN 10:52 I think we all have that with working out. IBBY 10:54 Yeah, I think with working out. I try to work out a couple of times a week. It's almost always an Alex Toussaint ride or a Emma. I love trying new instructors, especially as we've brought on a lot of new instructors over the course of the last couple of months. It's always super fun. But yeah, I am not a runner. I wish I was a runner. Unfortunately, I don't know. I'd never got into it. But I love the bike. I love the bike. I got one for my family. I have one myself. And it's a really good way of-- a lot of my friends have it now, too, and so competing with each other, making sure that if we have a ride that we absolutely love, we'll have a text chain where it's like, "Hey, this is a ride that we really--" Our teams, too, actually. Internally, at Peloton, we have a team ride that we do every Friday, which is super fun. My coworkers are just so much more in shape than I am. It's hilarious. It's just so hilarious. Oh, you know what, I'll be like, "Oh--" I'm like, "I'm doing well. I'm doing well for myself. I'm doing my own PR." And then it's just like I'm the bottom of the pack. SUSAN 11:52 Oh, that's so funny. IBBY 11:54 It's really funny. SUSAN 11:56 That's cool. I love the idea that all of your friends' data and your family's data and your team's data and everything, it's all wrapped up in your work too. It's kind of cool in itself. So that's a lot of fun. IBBY 12:06 Yeah, yeah, we definitely, definitely privatize that sort of stuff. So I can't tell who's who. But yeah. SUSAN 12:10 Oh, for sure. Yeah, yeah, awesome. So I'd love to hear maybe about one or two data science projects that you've done at Peloton that you're maybe especially excited about or proud of that you can share publicly? IBBY 12:23 Yeah, sure. I think the two most interesting ones would be-- the first would be furthering sort of our AutoML capabilities. So going back to what I said before, we want to really make sure that we democratize access to data and democratize building projects on top of data. And one of those things we believe is sort of operations machine learning and operations data science. So making sure that somebody who has a really, really good understanding of the business can do, quote-unquote, "feature engineering" without necessarily needing to have a Python stack. There's actually a lot of companies that are coming out these days that are doing this. [inaudible] is one that is really, really cool. But their entire mission, and this is something that we've built internally, is to take people who are very good at SQL, very good at structuring datasets together-- it can be, honestly, somebody who just uses Looker, Alteryx, or Excel, or a tool that can sort of aggregate data into a dataset. People who understand the business and the data for the business really, really well can do effectively the feature engineering part of data science and then quickly build propensity models, quickly build classification models, quickly build some of the-- even with time-series data, quickly build sort of a regression model. That's, I think, one of the most important projects that has found success for us where we've gotten the chance to-- a lot of folks at the company have stepped up and started using stuff like that, whether that be AutoML-- the other one is sort of a pseudo-related example, but it's auto research effectively, so using the same principles to understand the changes between different cohorts. So an example of this would be sort of templatized notebooks or templates, basically, where folks can come in and-- let's say that they're looking at two user groups, right? They're like, "All right. Here's a user that bought a Peloton bike a couple of years ago. Here's a user that about a Peloton bike more recently. How is the usage pattern different? What kinds of classes are they enjoying more?" IBBY 14:20 Yeah, so we've gone in, and we've created sort of this methodology to use automatic research-- to do automatic research where, let's say, you're trying to look at the differences between two cohorts, right, one that bought a Peloton or a user that had their Peloton journey start quite early, whether that be 2016, 2017, or someone who had started a little bit later, somebody who joined during the pandemic, "What sort of classes are they doing?" right, "Are they taking advantage of some of our newer offerings? Do they only take cycling classes? Are they taking meditation classes?" It really helps us understand, "All right. Where should we be developing new class types? What are people enjoying? How do we sort of predict what type of content folks are going to be relating to?" and basically give folks at the company a way to define two cohorts or define a test or do an observational study with internal data and understand very, very quickly how two groups of users-- or a study differs. So that's been another sort of infrastructure thing that we've implemented that has been very, very, very interesting. It just takes data from all of the different disparate data sources that we have, collects them all into one location, and then does a sort of visualization-- has a visualization element on top of it that other-- we'll do a bar chart or histogram or sort of a network analysis on different types of users. So it gives us a good way of figuring out where to sort of build next. SUSAN 15:46 Awesome. Yeah, yeah, definitely. No, that sounds very cool. And I'm curious, with both the AutoML and automatic research projects, when you've got people across the organization potentially using those, how do you get people to buy into those, to learn how to use them, to feel comfortable with them? How has that process worked for you? IBBY 16:08 Yeah, training is absolutely one of the most important things that you can do, right, trying to make sure that you set up technologies that folks will inherently use but also that there's a very, very limited amount of switching cost from the way that they're currently working on it to using the new system. And I think the best way is-- we just have a library of internal videos that we use to describe and show off what we're building. We have an internal channel where when somebody is building something new, they post about it; they say, "Hey, this is what I've built. This is sort of where it fits into the larger Peloton ecosystem," whether that be internal or external, "And here's sort of a brief video going over what it does." And so we do that in a little bit of a longer fashion. A lot of our sort of onboarding for new employees has a lot of these sorts of videos for tooling and a lot of documentation to where when somebody comes in, they can quickly go, "I'm trying to use Alteryx," or, "I'm trying to query something from Redshift. I'm trying to understand our data models." They sort of have a series of videos and documentation that they can use to get to that. SUSAN 17:15 Nice. Pretty cool. It sounds like a lot of effort going into creating all of those materials but also a lot of potential payoff for having people have immediate access to that. IBBY 17:22 Yeah, it makes it so you end up letting people answer questions very, very quickly and effectively, and there isn't necessarily a time sink from somebody else who's been at the company longer. It's perspective, right, so they're not-- obviously, we promote working together. We promote a lot of collaboration. But if it's a simple question, we want you to be able to answer it yourself. That's a really, really important part of getting things done quickly. SUSAN 17:48 Yeah, absolutely. And especially with, as we were just talking about, the incredible growth and expansion that you were dealing with, too. So offloading some of that onto individuals would be definitely a help in that process. Cool. So I noticed also looking through your resume and records out there that you were also teaching data science until pretty recently at Columbia. So I'm just curious what that experience was like for you and if that informed your day-to-day work at all, had any other impacts on your data science professional endeavors. IBBY 18:17 Yeah, yeah, I wasn't technically a Columbia employee or anything. But Columbia actually has a data science boot camp that they offer. They had to do with a partnership with the company. And I was a TA for a couple of cohorts, and then I got a privilege of a lot of teaching time last winter. And honestly, for folks who work in the data space teaching and for those who have been there longer than I have, you can probably get an adjunct role somewhere, teach different groups of students at different parts of their career, whatever. It's a great way to sharpen your skills and sort of help excite a new generation of data practitioners on sort of what's coming down the pipeline, right? If you're a part of the academic circles, or you're involved in boot camps, they're usually using very, very recent technology. A lot of folks either would read papers, or they'll understand, "Hey, this is a rising technology that's coming through the pipeline. We should develop our curriculum around it." And so one thing that was great was it allowed me to learn, honestly, a little bit more about the different types of data stacks that are at different types of organizations, right? So whether you're working sort of farther upstream; you're working in sort of the cache, the Redis world, whether you're working on the E and L part of ELT, rather than just the transformation part, and sort of helping students find what interests them within that sort of data stack world and saying, "Oh, this is the type of job that you might really love," whether that be machine learning engineer or data scientist, right, so people who are more on the analysis and model development side of things, or a data engineer or an analytics engineer or a data analyst or folks that want to end up doing product or visualization-related things, you do everything from Mongo and unstructured, no-SQL databases all the way to D3 and JavaScript implementations, which is really, really fun. It's a great way to sharpen your skills. It's a great way to excite people. It's a great way to honestly share your knowledge. SUSAN 20:14 Nice. Yeah, that's great. And then certainly, I've experienced that as well that as soon as you can-- what's the saying? If you can teach it, you actually understand it, something like that. And so yeah, that's cool. So-- IBBY 20:24 Yeah, yeah, what's really fun, sorry, was to go back and understand a lot of the math around sort of deep learning, around some of the certain machine learning algorithms because you learn those in undergrad, right? I had a class where we just had to implement a lot of the very basic machine learning algorithms on paper. But then when I started my job and when I started working in it, a lot of it's been abstracted, and you sort of remember, "Oh, my goodness, this is exactly how this works. This is sort of the process--" There's a recursive algorithm somewhere in here that you just run over and over and over again until you get to the result. And it's really fun to realize that the next day, you have to teach a class on it, go back, read through the textbook that you haven't picked up since undergrad, and go, "Oh, yeah, I remember exactly how this works," and show folks that. There's obviously a lot of folks that work in data, right, whether that be on a research team at a big company or whether they're doing deep tech stuff at a very small company that they work with the fundamental unabstracted layer of it every single day. But a lot of times, you just sort of don't get that when you're doing the more engineering side of things. SUSAN 21:34 Right. Right. Yeah, pretty cool. No, it sounds like a neat experience. So you mentioned that the boot camp was teaching some of the kind of cutting-edge stuff that's coming up right now in the field. What are some things that you're excited about in the future of data science, some things that really intrigue you that you would like to explore further? IBBY 21:55 Yeah, I think the biggest benefits that we see and sort of my thesis on where this is going to go is we're going to end up abstracting, I think, a lot of the development and deployment aspects, so the more computer-science elements of data science, into disparate sort of tooling, right? So you see this with the rise of things like Kubeflow. There's a lot of open-source projects that are designed to sort of take the development that data scientists do and then either interpolate the code and create microservices out of it to where you can actually use it in development where-- I think Alteryx actually has a pretty cool set of tools where you can run workflows that are around data science or machine learning and sort of abstracting a lot of the work that data scientists do because a data scientist fundamentally just isn't an operational engineer a lot of the time where when you hire a data scientist, you want to make sure that you're hiring someone who understands feature engineering, who understands hyperparameter tuning. You want to find somebody who has a fundamental knowledge of all of the different types of algorithms and what to use for a specific type of question. What you don't necessarily always find is somebody who understands how to deploy that somewhere, right? And so there's obviously the rise of MLOps folks. But I think, especially for those younger teams, having tooling that will help you deploy and use some of that modeling work that you're doing for research purposes in production is going to be a really, really important thing. SUSAN 23:24 Yeah, yeah, definitely. Cool. So I have one question that I always ask to our guests that we call the alternative hypothesis segment of the show. And the question is what is something that people often think is true about data science or about being a data scientist but that you in your experience have found to be incorrect? IBBY 23:44 I think the biggest-- I think the biggest fundamental thing that most of my students actually, when I was teaching, learned is that after they graduated from the boot camp, and they later went on and started their careers in data science, or they took their careers in one part of their company and moved it over to being more data-driven is-- one of the most important things that an organization can do is develop a really, really, really solid BI layer that-- I think that most data scientists come in, and they say, "Oh, I'm going to be doing a ton of deep development," right? You're going to find a problem. You're going to develop a machine learning model for it. And you're going to, A, set it, and it's going to solve all of the problems you've ever seen. A, that's not really how it works. You have to go back a lot of the time. You have to make sure you're retraining your models; you're looking at a lot of the fundamental assumptions that you're building off of. And a lot of the work that you're going to end up doing, at least when you start, is going to be a lot of solving and answering really low-hanging-fruit problems. And I think the biggest thing that I hear from data scientists that have worked for a couple of years now is, "I really wish I could get away from having to write really, really simple queries. I really wish I could get away from doing some of the more sort of lower-hanging fruit sort of work, so I can focus on making a bigger impact." IBBY 25:01 And the best way to do that, in my opinion, is to build a really, really solid BI foundation, making it so business users at your company can go in and say, "All right. I have the necessary tooling to be able to answer this question. I know where to go to answer this question. I know where the data lives. I'm able to sort of figure out all of the different types of scenarios that could sort of push this in one direction or another." And yeah, I think a solid BI layer and building automations is really, really important for that, right? Whoo-hoo, Alteryx. Just like the ATM, right, it revolutionized getting money and sort of left bankers' time to do more important things. I think that you can sort of view an automation and business intelligence layer as a way to sort of understand bigger-picture things around the company, especially when you're small, right, because a data scientist at a young company is going to be doing a lot of wearing every type of hat, right? That's something that you probably hear all the time where, "I'm a data scientist and we're small. I wear a lot of hats. I do everything from the E and L side of ETL all the way to deploying deep learning models in the cloud that we're going to use on our app side." And I think the biggest way to be able to fundamentally move forward as a data org is to make sure that the first thing that you're taking care of is having a really, really solid analytics layer to build off of. SUSAN 26:20 What do you think are some of the barriers or challenges to making that happen? IBBY 26:26 I think one of the biggest ones is just stack knowledge, right? I think a lot of what ends up happening is you join a job; you're focused on building out a very, very specific set of tooling, or you're focused on solving a very, very specific set of problems; and then you move somewhere else where you're doing a lot of the same sort of stuff. And I think parochialism is sort of the enemy of collaboration, right, where what ends up happening is we have either a lot of experts in a specific thing that don't necessarily always talk to each other, or we don't have someone who understands at least a small element of all of the different types of decisions that a business has to make to where they can connect all of the data together behind the scenes and make a really intuitive layer to help people understand that. SUSAN 27:15 Yeah, great points. Is there anything that we haven't talked about yet that you would like to get in there? IBBY 27:20 I think for anyone who's listening, I'd love to pub that if you're interested in chatting-- one of the things that I love the most is helping folks who are building new data products, who are thinking about the data paradigm in a completely different way. So I mean, honestly, an open shout to all of your listeners is if you're building something interesting, if there's something that we can chat about where you have an idea, we love meeting entrepreneurs that are building interesting things, so we can onboard their products at Peloton. I love just geeking out over different ideas. And so if you're out there and listening, and you're like, "Oh, I'm working on something," or, "I think that this paradigm that we're using the data world is wrong and that we should change it to be something else," those types of people are always the most interesting to talk to, so feel free to hit me up. SUSAN 28:04 Awesome. Well, thank you for that. I know people appreciate that opportunity. Very cool. Well, Ibby, thank you so much for taking the time to talk with us today. I really appreciate it. And I know that folks are going to have a lot of interest in the things that you've talked about, and it will give them a little more insight into their Peloton experience, too, which is pretty cool. IBBY 28:19 Yeah, yeah, thanks so much for having me on. SUSAN 28:22 Thanks for listening to our Data Science Mixer chat with Ibby Syed. Join us on the Alteryx community for this week's cocktail conversation to share your thoughts. Thinking about Ibby's experience teaching, if you have the opportunity to guest lecture a group of budding data scientists on just one topic, something you think everyone getting into the field should know and understand, what would that topic be? Why do you think that's so important? Share your thoughts and ideas by leaving a comment directly on the episode page at community.alteryx.com/podcast or post on social media with the hashtag #datasciencemixer and tag Alteryx. Cheers. So one thing that I forgot because we were talking about the record button is I forgot at the very beginning to ask a very important question, which is that on Data Science Mixer, we typically have some sort of snack or drink or something with us while we're recording, so do you have anything special there with you today? IBBY 29:23 I have a water, so I could stay-- SUSAN 29:27 That's a popular choice. Yeah, we got to be ready for all that cycling, so I assumed-- IBBY 29:32 Exactly, right? I feel like I've been trying to be more healthy, trying to load on sugary drinks. However, my usual go-tos are mango juice during the day or a yogurt drink. I just had a quinoa salad right before I joined-- right before I joined the call. That was sort of my lunch, which was really, really delicious. Yeah, no, today is just boring, no interesting drinks today. I knew that you were going to ask me this question, and I knew that I should bring something up. And then I went to my fridge, and there were two options. One was milk and the other one, water. And I was like-- SUSAN 30:06 It's all good. IBBY 30:07 "--I'm just going to go with the simple water," because if somebody asks me what interesting thing I'm drinking, and I say milk, I'm going to get judged, I think. I'm not super worried about getting judged, but I am a little worried about getting judged. SUSAN 30:20 Gotcha. No, no judgment here. And water makes a lot of sense. It's been a very popular choice lately, actually. People are apparently actually trying to work during the workday, so we do these podcast recordings, and they're not imbibing, I mean, really. IBBY 30:33 Ugh, what? How? Why? Why am I not imbibing? It's a Friday. Why am I not imbibing? No, you're right. You're right. SUSAN 30:38 That's true. That's true. And it's a little later over there. No, I just have some good old black coffee here, some nice French roast, so, much needed getting through-- IBBY 30:46 Oh, that sounds good. SUSAN 30:47 Yeah, yeah, good stuff. IBBY 30:49 Do you grind your coffee yourself and do the entire process, or? SUSAN 30:54 No, as somebody who lives in Oregon, I have a really unsatisfactory answer to that question, which is I buy pre-ground French roast at the grocery store. IBBY 31:02 I do too. I do too. Trader Joe's French roast is one of my favorite-- one of my favorite coffees, so yeah, very good. SUSAN 31:07 Yeah, it's good stuff, for sure, for sure. Yep. Yay.

This episode of Data Science Mixer was produced by Susan Currie Sivek (@SusanCS) and Maddie Johannsen (@MaddieJ).
Special thanks to Ian Stonehouse for the theme music track, and @TaraM for our album artwork.

Data Science Mixer

Episode Guide

Leading the data science pack at Peloton | Ibby Syed

Panelists

Topics

Cocktail Conversation

Transcript