Resources
Sep 15, 2021 |
1h 8min
|
1.8k views
|

Data Science Hangout | Sep Dadsetan at ConcertAI | Infrastructure that Encourages Reproducibility

We want to help data science leaders become better.

The Data Science Hangout is a weekly, free-to-join open conversation for current and aspiring data science leaders.

An accomplished leader in the space will join us each week and answer whatever questions the audience may have.

We were recently joined by Sep Dadsetan, Executive Director, RWE Analytics at ConcertAI.

01:35 - Start of session
08:20 - Gaps in the data science space
12:13 - How to show key stakeholders the value of data science
23:23 - Setting up the infrastructure to encourage reproducibility
29:38 - The importance of architecture and infrastructure to data quality
35:34 - Adoption to formalized processes over individual processes

► Subscribe to Our Channel Here: https://bit.ly/2TzgcOu

Add the Data Science Hangout to my weekly calendar: https://www.addevent.com/event/Qv9211919

Follow Us Here:
Website: https://www.rstudio.com
LinkedIn:https://www.linkedin.com/company/rstu...
Twitter: https://twitter.com/rstudio

image: thumbnail.jpg

Transcript#

This transcript was generated automatically and may contain errors.

Welcome to the Data Science Hangout. Thanks, Sep, for jumping on and co-hosting with me today. So welcome back to all the familiar faces and to those joining for the first time. It's really fun to see this like recurring group starting to form. But for anyone new today, these discussions cover a variety of different topics on data science leadership in the enterprise, but really focus on questions that are most important to you all. So again, no agenda and everyone is welcome to join in live or put any questions that you have in the chat.

That said, I'd like to just jump in here and introduce Sep Dadsetan, Executive Director at RWE. And I am checking with you that is real world evidence analytics at ConcertAI. But Sep's passion about learning and exploring new concepts and ideas and, and bringing them to life. I found that on your website somewhere. And I will say, Sep, I did learn a few random things about you making that into like DJing and footgolf. But would you be able to kick things off by introducing yourself and sharing a bit about the work you do?

Yeah, happy to. I'm glad to be here. I think I've already been in part of several weeks of discussion and it's highly valuable. So I definitely encourage you guys to share your perspectives. I think this is a great opportunity to just learn from everyone. So yeah, so my name is Sep Dadsetan. My formal training education is in actually like molecular physiology. So I studied immune cells and T-cells. And then made a, did a brief postdoc at Genentech studying B-cells, another immune cell. And then that's when I said I was going to leave academics and join a startup company, which was CyApps, where we worked with real world data and real world evidence there. So that was most of my tenure, seven and a half years there. Brief stint at Kite for a couple of years doing IT for their R&D group. And then recently joined ConcertAI back in the real world data space about six months ago now.

So that's kind of the brief tour there. With respect to what I do, at Concert, I'm mainly kind of focused on achieving kind of scalability, reproducibility, doing some training, and trying to, at this point, really kind of focusing on architecture and trying to kind of gather our teams and smooth some edges, so to say.

Awesome. Thank you, Sep. So while we kind of wait for people's questions to come in, I'd love to kick it off with one of our questions we often ask people. But what is something that you're really excited about in data science right now?

Excitement around data adoption

I think broadly speaking, I'm just excited about the level of adoption businesses are having around data in general and how it can impact their business. So I remember probably early 2010s, data science is this kind of term that's popping up. People are kind of, what is it? What is it? It's probably still today we ask what that is, right? But a lot of businesses really weren't really sure what that meant and how it could benefit them. But I think now, like a decade later, we're seeing a lot of companies IPO-ing that are in the data space. We're seeing increased adoption of cloud computing. We're seeing a lot of businesses really shift towards data, which really kind of, I think, solidifies that data is here to stay. All data science, data analytics, and data-adjacent roles are here to stay and that there's a lot of value to be provided. So that's, to me, exciting.

And we're still kind of obviously shaking a lot of those things out to figure out what that means on a broad level. Yeah, I think what's interesting probably on the forefront is also synthetic data. I haven't played a whole lot in that space, but that's been pretty interesting to see how, yeah, we generate a lot of data in general, but not a lot of that data is necessarily usable. Or if it is usable, it may be sensitive, in which case we have different strategies to allow technologies like AI and ML to kind of proceed in a more safe manner. So that's kind of exciting.

Awesome. Is a company being data-first something that you first look at when deciding where you wanted to work? Or I was just talking with someone yesterday and they were saying when they were deciding which companies to take a look at, that was a huge factor for them and maybe something for people to consider looking for new roles.

Yeah. I mean, yes and no. For me, I like a challenge. So sometimes they want to be data-first. And so that's where I'll raise my hand. So cool. Let's figure out what that game plan is going to be. But yeah, I mean, obviously, I think a business and especially the leadership to have the interest to want to go into that direction and enable, either enable the company to be that way if they're not, or to just enable the company to continue to be that way, I think are both highly important attributes, at least for me, and I'm sure a lot of the other people here on the call that are so interested in data.

Gaps in the data science space

Definitely. I'm waiting for a few people to put some questions into the chat or feel free to jump in live. But yeah, something else I thought might be helpful to cover is certain challenges that you face or something that you think is broken in data science today.

That's a great question. There are a couple levels to that. So I think I'll start at, let's say the lowest level, which is the analyst data scientist. I think that since there isn't a single path to become a data scientist, people come from all walks of life, you also then get this varied training. And it's like a pro and a con, because on one hand, usually they have domain expertise, usually they have, you know, great skill set, but those with teams or how they've worked with teams in the past.

So I think one of the interesting aspects or I guess gaps might be how people come into data science and the varied, you know, directions that they come, which would kind of, you know, when we talk about training and like trying to get everybody on the same page, it's a little bit more difficult than if you were to kind of kind of, let's say, have a standard path to become a data scientist, so to say. So I think that's one of the challenges currently in the field. At the higher level, I think we're in a bit of a transition period as businesses are starting to kind of realize the value that that data might provide them. Not all, not all, let's say leadership or a lot of businesses are data savvy. And I think, again, that goes back to this training aspect where, you know, you need to translate your work for a business leader or someone who might not kind of understand it. You know, that still exists and it's kind of will always shape how, let's say, the data science groups get resources and funding. And it will also shape, you know, how businesses will kind of move forward and plot their course. So those are the two kind of, I guess, levels that I think are gaps in our current data realm.

Showing stakeholders the value of data science

Yeah. On that point of kind of communicating it out to the business, I know, Frank, you just had a question around business impact. Would you want to ask that one live?

Sure. So everyone has a different answer to this, but in terms of business impact, how do we show stakeholders or users the value it brings? Obviously, it's making its way to the top and leaders are acknowledging it because to your point, Sep, this industry is growing. The jobs are here to stay. Are there any examples that you have on how you demonstrate that? Obviously, the more quantitative, the better, but oftentimes that is a challenge.

And so to, I guess, to clear that up, so are you referencing, like, let's say, enabling a new arm of the business or just to show the business that, hey, look, we need more resources. This data is actually very valuable and this is how it will.

Yeah, I guess I would go, I'll skew it to my perspective, but when I think about how stakeholders and leaders view the data science teams and folks, I think they think that every data scientist can do everything, all of the tools, like they can do everything, right? So they're like, you have five people, right? You should be able to do all of these projects and maybe even more. But the truth is there's only one person who has experience in forecasting, right? And it's super challenging to take on the 10 forecasting projects for the one person and get the other people up to speed. And then at that point, they're like, well, I mean, if you can't do that, then you cost too much until you can offset that with what is the value you're showing up with, whether it's like cadence of delivery, how fast you can work. And it's really hard, I find, to relate that all the way end, all the way to the end of the cycle where you're actually saving business costs or bringing in increased revenues.

Yeah, I mean, that's a great question. And I think that that generally applies to not only data science, many other teams as well. I mean, even engineering teams, sometimes you have one person that's like the front end developer, right?

You know, I think that what I've found success with, you know, and it's hit or miss sometimes, right? And it's not everything is a single solution or single path to a solution. What I found success is just trying to be as close to the business as possible and understanding what their needs are. Sometimes you find that they just move too quickly to even know what their needs are. And so you have to kind of help guide them. And in doing that, you build this kind of partnership where you then have more value and understanding of how you can kind of guide, you know, orient your team to provide that value. And then they get closer to understanding it.

Um, that would be kind of how I've found a bit of success. But yeah, it's not always it's super, it's not always straightforward. Is there anybody else that also has kind of want to chime in as experience in that?

Yeah. And I mean, you know, like, I'm trying to think of, think of examples. I mean, sometimes it's so, so my experience has been very, it's been quite varied. I've worked as like early I've worked at like large pharma companies, like, like, you know, Gilead kite, right. And it always varies right at a small startup. You know, it's always go, go, go. Priorities are shifting all the time. And so where I ended up trying to provide value was, yeah, occasionally you have to make those deliveries as, as requested. But what I've found is that, you know, it's always going to come back to, you know, you have to make those deliveries as, as requested. But when you're building the infrastructure in a way that makes that delivery, let's say reproducible or makes that infrastructure reproducible. And then, and they see the speed at which you can turn things around in the future. Because inevitably, we all probably experienced this. Someone's going to ask a question or want a delivery. And then they want some variant of that delivery three months from now, right? And so you go build, you build that first iteration, maybe it takes you, you know, a week longer to do it. But if you do it right, the next time they come around, it turns around a day, they they fall out of their chairs, right? And I think that that's where all of a sudden, you're you're showing that, look, this is this is the value of doing things in a manner that's that's reproducible and scalable, because we're saving costs, where we are turning things around that impresses, you know, the customer. And, you know, and those are the things where you're like, okay, you know, regardless of who's doing it, that the operation of the data program is, is efficient, right? And so I take I take pride in that I love kind of efficiency in that regard. And so kind of scale and all those things are kind of an interest for me, but I don't know, now I'm just probably blabbering, I'd love to hear other people's take.

And I think that that's where all of a sudden, you're you're showing that, look, this is this is the value of doing things in a manner that's that's reproducible and scalable, because we're saving costs, where we are turning things around that impresses, you know, the customer.

How much how much time is spent on the prep work, I guess, for doing what most people would think of, like, you know, the data science analysis versus, versus doing that makes sense? Oh, yeah, yeah. I mean, this is I think it varies by industry. In the industry that I'm in, I mean, we're dealing with with people's health data. And it can be very, very messy. But at the same time, it's very, very important. We're trying to recreate a patient's journey through treatment and whatnot, and we want to make sure that we get it right. And so it ends up being quite slow, quite methodical, there's a lot of people that are involved to make sure that we're making clinically appropriate assumptions to then finally get to, you know, how we get to our data set, right. And so it's probably a lot longer in the real world data space than than perhaps others, but I don't have a whole lot of experience in, in other other types of data sets.

I think, you know, you probably the adage that's like 80% of a data scientist's time is spent cleaning, cleaning data and organizing data might be appropriate. I don't know, I've never really kind of thought about it. To be honest, I just kind of like get it done and try to do it in a manner that I don't have to do it. Like repetitively try to build on knowledge. And is that something that you have to communicate with other folks in the business about or do they usually understand? I know that there's a lot that goes into, you know, the pre-cooking.

Yeah, I mean, I think that goes to, I guess, to go back to Rachel's question about what gets me excited, right? So what gets me excited is that great, all these businesses are very interested in data, we're here to stay, right? But there's a flip side to that is that data literacy also is required. And so there are various levels. So some people are happy to understand the nuance and the nitty gritty of why it takes so long to kind of stitch these things together. And then there are others that are saying, look, we need to deliver to the customer, this is what we want. And I get that it takes time, but hurry up, right? And so, yeah, I mean, I always try to kind of simplify and really kind of point out, give status updates. I mean, transparency for me is really important. So if you're transparent with your stakeholders and they recognize what the blockages are, and I think everybody that you're going to get engaged with is going to want to know something that you might be able to help out. I think that again, helps with establishing that relationship and having them get a little bit closer to the data and understand some of the hurdles that you might face.

But yeah, sometimes patience is perhaps thin and they don't quite, quite get it. But that comes with this other side. If you're going to be more data company, we're going to have to have more data literacy and what are better methods of perhaps working with data and how it can enhance us rather than, let's say, throw people at a problem, do it intelligently from an engineering standpoint, so that you can scale your operations. So you can throw more people at it, but when you're throwing people at it, you're providing more company value rather than trying to wade through the mud with increasingly large data sets or increasingly complex data sets.

Scaling relationships and partnerships

Hey, I'm going to jump in. Sep, you mentioned, I think you used two words, partnership and relationship somewhere as you were describing that. And I have found every single time I've had a new team that partnership and relationship with my stakeholders and the business users is immense, right? Once they trust you and they say, oh, you're actually on my team, like we're working towards the same goal and they get it. And one of the tried and true ways I've done that is if it's transportation, I read a transportation textbook and we're not in the office anymore. But once my partners see that I have a transportation textbook on my desk, they're like, oh, you really are on this team trying to solve this problem, right? You're not just the data person or data team over there. That's huge. With that said, I've had success there, but scaling relationships and partnerships is really hard because they both take time, right? And we want to be data teams and we want to do things efficiently and scale it across hundreds of thousands of decisions. I don't know, does anyone have any thoughts on how to balance scale and efficiency with relationships and partnerships?

Yeah, I'd love to hear from others too. That's a great point. Hi. Hi, everyone. Hi, Sergio.

Hi, I'm Sergio and I'm actually connected from Bolivia and it's so interesting to be here and listening to all the experiences you're talking about. And actually, well, I started a company, a data science company here in Bolivia about maybe six months ago or something like that. Well, previously, me and my partner, we have been working on different consulting projects related to data science as well, managing data. But it's so interesting right now that, well, I want to share you what my experience is at this point. We believe we are doing really well because we're starting to get involved in interesting projects with different sectors, for example, mining sector, financial sector. And, well, what we have noticed and, well, our approach is to or our strategy is to have our first meeting, for example, with these different companies. And we are not offering exactly like solutions or services or products. Like our first approach is to try to make them understand what all this idea is about. And, well, something that is super important, I believe, that they have to understand is that, well, data science is kind of a structure that is, well, it has three main parts. That is basically, well, the programming part, the IT part, then the statistical and analytical part. And then, well, one of the most important is the know-how and how everything works.

Actually, what Frank was talking about, like talking about transportation, for example, being a partner or working for them, you have to start understanding what the work is about and understand being part of the company. So I think that's super important. Like when you try to have a, well, a relation with any company or any industry, first you also have to understand what it is about and how it works, how everything works and work really, really close to them to be able to develop these solutions. And, well, that's the experience we are living right now.

Setting up infrastructure for reproducibility

It seems like the, like when you talked about infrastructure, like I think sometimes you think servers, but you were talking about like how you actually set up those projects so that you can make things reproducible. But it seems like those relationships are almost a part of the infrastructure as well. And it's curious, just talking a bit more about how you do make that infrastructure reproducible, Seb. Yeah, great question. So yes, I mean, reproducibility and scalability all, I mean, it relies on the success of training, making sure everybody's doing things and the processes are all the same. And it also is related to this partnership and relationship.

So I think it's interesting for myself that I spent a couple of years in IT at Kite Pharma, which is a common thing that, you know, a lot of people on the other side of the analytics side need to deal with at larger companies. And so it was really nice to kind of get the perspective from the IT side and now come to concert and say, okay, I get how it works. Like I can, you know, so yeah, so part of reproducibility for me is, you know, everything, I mean, it's like, it really involves everything. So, you know, making sure everyone's kind of using, you know, let's say version control the same way, that naming conventions are the same. We built a, basically a package, an R package so that when people create new, because we use RStudio. So we create a new package. So when you create a new RStudio project, that project will always be the same. So it creates the same folders, the same setup. So when I'm talking infrastructure to make it reproducible, everything from how you use code, how you name your code, how you name your variables, how your projects are set up, how your files are named, all of that stuff. They're really, really annoying little things, but they bubble up to be so much more because when you create a project, for example, an analytical project, everybody that it becomes a transposable unit. That means I can take my project. I go on vacation for a year, someone emergency request, someone else can pick it up and know exactly where the data is, know exactly where the scripts are, you know, et cetera. And they could, they should be able to pick it up with a lot less, you know, overhead for having to figure out, oh, well, how did SEP organize this stuff? Like, I don't know. Right. That, that takes sometimes days.

they bubble up to be so much more because when you create a project, for example, an analytical project, everybody that it becomes a transposable unit. That means I can take my project. I go on vacation for a year, someone emergency request, someone else can pick it up and know exactly where the data is, know exactly where the scripts are, you know, et cetera.

So that's one. I think a big part of it actually, to be honest, is a lot setting up a lot of that real basic stuff. And then it also makes it much smoother so that every time you set up a project, you don't have to like, think about, think about those things. You just kind of go, this is where you write your stuff and whatever. And then we happen to use RConnect as well. I know that there are other solutions, but that allows us a feedback loop. I think one of the aspects that, you know, that we ended up talking a lot about in these weeks are like communication and how you communicate with business stakeholders. And so, you know, if we're going to set up that, you know, RStudio project, we want to then be able to publish it, whatever that output is, whether it's an API, whether it's a product, a report, whatever, an application, we want to be able to publish it and publish it in a central location. So people aren't trying to search for it everywhere. And then we have a URL, we could share that URL and we get this feedback loop. It's not sitting in somebody's email and part of a chain. You have to then go through your email. And so we've kind of streamlined it. And that allows us to kind of, you know, basically hit a conveyor belt. If there's iterations, changes, we can kind of do that. And then it's modular so other people can kind of pick it up. So that's the kind of, I guess, reproducibility and scalability that I'm referring to.

Data quality

Thank you. And I see Trevor has a question which kind of touches upon this as well around quality of data. Trevor, would you want to introduce yourself and ask that question?

Hi everyone. Yeah. Nice meeting you all. Very good sessions every week. Yeah. So my name is Trevor Thompson actually from Jamaica. Work for a global gaming company. I'm very interested in the infrastructure. When we talk about data, we speak about models, we speak about data science, but the emphasis on talking about the quality of data that we use. Garbage in, garbage out, right? And the process of collecting, cleaning and storing. I know it may be it is a data engineering process, but of course it's very important when you talk about data science. Even since week, I was involved in a process with an application that collects the coordinates of some locations. And we found out recently that the collection process is not precise enough. So we're ending up in a situation where we have this consultant who wants a great degree of precision in terms of identifying a location, but the radius of the collection points is like 83 meters. So that's definitely not as precise as I would want it to be. So some of these things you don't even realize until you're really digging deep, right? So this has always been on my mind when we talk about the quality of data that we use all throughout the data science process. So it's just about thinking about that. So that's what's on my mind there.

Yeah. Thanks for chiming in Trevor. By the way, I love your accent, Jamaican accents. So feel free to chime in further. If anybody else has something to say about it, I know there's many ways architecture and infrastructure can refer to a variety of things. And there's many ways to kind of solve for whatever your business purposes are. I mean, generally you have to partner pretty closely with your data engineering team. It kind of depends. Sometimes data engineers themselves have a little bit of an analytical bent to them. So they're always kind of constantly checking. Other times there's a dedicated QC team that's constantly evaluating and kind of sampling as data moves from place to place, just to make sure that you are collecting the right information and that all the transformations are going accordingly. Other times, and again, depends on the size of the company and how roles are broken up. The data science team should have someone that is actually kind of evaluating or at least generating some sort of report out of

wherever the staging areas of your existing data is sitting to kind of hopefully prevent that where you do all this work and eventually you get it and you're like, oh boy, this is not what we wanted at all. So many ways to kind of solve that. I don't know what your particular setup is, but those are a couple of ways in which we've solved that in the past, which is just to kind of somehow, through whatever team, get our hands on things a little bit early and sample just to make sure things are kind of as we expect them to be.

Yeah. Anybody else have thoughts on that?

Cool. Trying to wait for like seven seconds for people to chime in there.

But yeah, if anyone else has thoughts about that, I think Mike Thomas, you chimed in a little bit about that and more so about the time spent on design. Do you want to add anything there?

Yeah, I don't know. I think, you know, Sep spoke pretty well to it. I just find that communicating upfront and I think design kind of does this intrinsically is when you spend a lot of time on designing kind of what the output solution is going to be. It requires you to bring all the stakeholders to the table upfront, you know, the owners of the data, the owners of the platform, the people who are going to be using your solution, you know, on the ground day in and day out, the technical folks in your analytics team, and then also the IT team that might be enabling the solution to even work, security, infrastructure, stuff like that. I think when you focus, when at least you attack the project with a design mindset up front and really focus on nailing down, you know, what the end-to-end process is going to look like, I think it pays off quite a bit in the end in ensuring that, you know, whatever you put together is well architected, you know, maybe you're ensuring that there's a data dictionary behind all the data that's being used, you know, things like that, and that the output is what everybody's expecting it to be.

I find myself nowadays probably drawing as many, spending as much time drawing architecture diagrams as I am writing our code, which, you know, is fun if you're into that, but I think it's an important step to take in any project, and I think it's a good team mindset to have as well.

it's an important step to take in any project, and I think it's a good team mindset to have as well.

Balancing formality and speed

I like Daniel's question from earlier that kind of relates to this as well, Mike, but like balancing that form, like making the process and making it more formal versus getting the results quickly for someone. Daniel, would you want to kind of chime in and give your background and how you've experienced that? Sure, sure, absolutely. So my name is Daniel, I'm a Lead Data Scientist with Nestle, and it's touching a little bit on what Seph you were speaking to before, in that kind of like if time was not a requirement, then I think, I personally know a lot of people in our positions would kind of want things to be as formal and as repeatable as possible, so that to your point, like somebody else can come in if I'm away, or I can easily understand and read somebody else's code, but the drawback is that takes time to do, and so one of the things that I struggle with personally on the team is I probably lean much more towards the formal side, and not everybody does, and I have struggle figuring out how much to kind of push those others to kind of fit my system, so to speak, at the same time knowing that they're highly competent data scientists in their own right, they just structure things a little bit differently than I do, and if I kind of not force them, but kind of try to influence them to fit into one standard way of doing things, then it slows them down, and the end result of whatever it is that they're building or analyzing isn't as available as soon as possible, so it's just that constant struggle at Nestle now, but as well as other places where I've been, like how much do you try to make things repeatable and formal versus kind of like every person is their own system?

Yeah, that's a great question, and it is a common situation for better for worse, again, it depends on the culture of the team and what the business is really kind of after, I think, and it depends on, I don't know, what relationships are like, I like to try and lead by example, so at the very least, if I'm finding that it's difficult to kind of, you know, hey, like look, this is a large team, or the team's really, everyone's kind of just doing their thing and in their own lanes, at the very least, I try to simplify it for myself and try to make my work more reproducible, and in many cases, what ends up happening is whether there's, you know, code review or whether there is, you know, future fire drills or whatever, that process will always win out, right, and when someone else sees it, they're like, oh, maybe that's cool, maybe I want to try that, how did you do that, right, and so, yeah, sometimes it's not so easy just to kind of, you know, go dictate to people and say, look, like this is how you should do it, until they kind of come to the realization themselves.

And so that was, you know, that's one method that might work, where it's just, at the very least, you could do it for yourself, it does take time, but even small wins are important, right, so I've been in situations where, I mean, especially in a startup, where it's like, okay, we need something in two weeks that we don't have anything, like nothing exists right now, but we need to deliver in two weeks, right, so it just ends up being, okay, just dive onto the keyboard, but even in those cases, wherever I can have kind of an engineering mindset and modularize what I can, I know that those pieces can be used either on other projects, or this project becomes much easier with future revisions, and I think that goes back to what I was saying, is that everybody comes from so many different angles into data science, sometimes that engineering piece isn't there, and so, you know, showing people that you could do it a particular way might be helpful.

I might have a couple things to add there as well, if that's okay. Yeah, of course. I think probably, I think this is what Seth was speaking to, but maybe this is a more high-level view of the answer to your question. I think it's a belief that collaboration in the end will make your team more successful, and make your projects more successful and more efficient, right, you're not going to get collaboration, collaboration is not going to go as well if everybody's in their lane, and they have different standards, and you don't have necessarily code review, because somebody's writing their code in one style, and other people are writing their code in a totally different style, so I think the way maybe to go about fostering that, or preaching it, would be to get people to buy into the idea that collaboration is going to make your team more efficient, effective, get you, in the end, at the end of the day, once you have things set up, you know, get you those results quicker, and I think Seth made a great point about small wins. I think small wins, in this case, in terms of, you know, just even styling functions, or documentation around your code, and things like that, can go a long way towards turning on the light bulb in people's mind about how, you know, how you're going to make your team more efficient, and turn on the light bulb in people's mind about how some standards around how folks are writing their code can be effective.

The value of infrastructure and communication

Yeah, so just to add a bit there, so one of the selling points for us when we talk about data science, and about using tools like R, is about efficiency, right? The ability to produce reproducible code, or results, and efficiency. So even though we may think documentation and structure is boring, and it is, we know if it's not in place what the results are, right? So essentially, something that will take you two minutes, may take you an hour, right? And in terms of even communication, communication with members on your team, or even outside, when there is no commonality, you're talking about, in Jamaica, we call it Chinese telephone, where you're trying to get to someone, you can't get to them because the language is different. So, you know, what is so interesting about this conversation now, is that we're talking about infrastructure, we're talking about communication, and none of this is really about data analysis itself. But how important all the tenets are, all the connecting points around the data analysis itself is extremely important. That's what I get from that.

we're talking about infrastructure, we're talking about communication, and none of this is really about data analysis itself. But how important all the tenets are, all the connecting points around the data analysis itself is extremely important.

Definitely. Trevor, I want to save that snippet that you just said there, because I think that is so, so important. And also, like the thinking about the communication back to the team about that as well, and like, how do you communicate that back to the business if, oh, they got this one report in one day, because you had already set that process up? How do you help them understand, like, every report doesn't just come in one day like that?

Yeah, yeah. I mean, it's, again, it's like showing the art of the possible. And I think when they realize, you know, you do things in a smart way, then you can gain speed. They're more than happy to then support that, right? So, if a new request comes that's outside of the scope of the original request, they then have an understanding of what it took to make that original one, and then they'll want it again. Because it's, again, it's like building knowledge, and it's a building scale, right? You're not going to get everything in one go, but if you do things, you know, majority the right way the first time, over time, you're going to have a bolus of knowledge that you're enabling the remainder of the business, whether that's other groups, whether that's marketing, whatever, to be able to do what they do. And I think that that's where the value comes out, right? It's not just, you know, sitting at a keyboard and, you know, writing out the script and getting a result. You could do that, and it works, but you're trying to think a bit more long-term. So, yeah, that's at least how we've approached it.

Building relationships and going on tour

Are there a few small organizational or collaboration activities that are sort of your go-to's, you know, when leading a team or, you know, group of people into a more formal way of doing things? I can't share that information. That's proprietary. Okay.

I've got a secret, no. No, I think, you know, I don't know. I think there's, you know, one of the questions that I think that was discussed perhaps in the past was, like, people that are interested into getting into leadership, right? There's naturally going to be a transition point where your hands aren't on the keyboard as often, you know, where you're not doing the analysis per se, but now you have to manage a team and grow that team. And with that come different responsibilities. And so, some of those responsibilities are just evangelizing your team and being able to communicate and work with other business partners. And so, I think, you know, I've been now in, I think, two situations where I have either started a team from scratch or the team is basically pretty new. And I think the most important thing is, obviously, you get your bearings first, but then basically going on a tour, right? Speaking to all the business partners, different stakeholders around the company and trying to understand what it is they do, how they engage with data, how they envision working with that team, and just maintaining those relationships, like constantly having conversations and not being afraid to, like, help out when you can, right? Sometimes some of these groups might be resource strapped. And so, jumping in and trying to solve little problems here and there may be helpful. And I think that builds the rapport. It kind of accelerates a little bit about this partnership building and, you know, really allows you as a group to have a better understanding of the business so that you can align your priorities in accordance with that and the goals of the business, but also allows them to understand how you operate and what benefit you can bring to the table. And then when those things kind of align, then you start to be a little bit more closely integrated.

Does that answer the question? And the one follow-up I have is, on that tour, is there, like, a balance between fielding requests or projects and then, like, seeking out areas where, you know, you want to bring something to the table?

Yes, and I think this depends, again, on how the business is led. So, if the business has very clear goals and guidance, then your group should always really kind of be aligned with that.

Other business units may or may not have, you know, alignment in that regard. And so, you can't just, I mean, some people are nice and they just want to field whatever request. Okay, fine, you could do that. You're still going to, you know, get some wins, but it'll be also a bit more distracting towards you achieving what you need to do as a team to become most efficient, because now you have, perhaps, extraneous projects. Whereas if the businesses, you know, some businesses might not be so clear on their objectives and goals, maybe it's new, maybe it's fluid. And so, that requires partnership with those other business units. Are we really, you know, doing the best, biggest bang for our buck in doing these projects?

And so, yeah, there's a balancing act. You don't want to field all requests. It doesn't hurt to have an understanding of all the requests, so that you have an idea and they can maybe say, hey, you know what, like there's a piece of this that we might be able to work on, or a piece of this, or maybe we can just tackle this for you. And then, and don't be afraid to kind of also ask, right, ask how other people might be able to help. Sometimes that's just, you know, being involved in maybe, you know, weekly, or bi-weekly, or bi-monthly, or whatever presentations that they have, so that you're aware of their work. And I think that that shows that you have an invested interest in them, their success, and hopefully they reciprocate as well. And I think that then brings the whole business together, and it also really helps the data science and data analytics group as well.

Awesome. I'm trying to like group together some of the questions, so that we don't keep like jumping back and forth. But I think on this topic, I saw Tracy, you commented on a helpful resource for cultivating collaboration in data science. Would you want to maybe introduce yourself and provide any additional context on that?

Sure. Hi, I'm Tracy Teal, and I'm the relatively new open source program manager at RStudio. And so, I'm focused on the open source tools that RStudio is building, like the Tidyverse, and the Elmoverse, and Shiny. What's Tidyverse? So Tidyverse is... I'm kidding, I'm kidding. You know, there's all kinds of different stakeholders. I'm first on this call. I was on a Tidy Models meeting, so I'm sorry, I was a little late for this one. But yeah, I'm just, you know, wanted to get a chance to hear, you know, how people are using tools, and what challenges they're facing.

And I think one of the themes I've seen throughout my career, I've worked in non-profit and academic spaces as well, is where the challenges are more social than tools, which has been, you know, a lot of this conversation. And so, I was part of a collaboration where we put together this sort of, there's this framework in PLOS of 10 simple rules, none of which are actually simple, but that's the title of the series. So we put together kind of the set of things to think about in these collaborations, because like what you were saying, Trevor, of the situation of like getting into a project, and expectations aren't the same, and how do you talk to each other? Those are always themes that come up in any project, and it feels like, oh my gosh, I'm the only one that's doing this wrong. But it actually is like one of the biggest challenges. So I've loved this conversation, that might have some guidelines there.

I think one of the other things we focused on in that, and paper, and we think about in our work and in open source as a whole, is really about the inclusivity of the environment, and for really a successful collaboration. So setting up that kind of framework in advance, really allowing everybody to participate, especially when you have these challenges of language that's not shared. Like your data scientists maybe use words that your domain researchers, or your domain experts don't. So a lot to learn on each side of the fence.

Definitely. Thank you, Tracy. And Tracy, I see you shared that link a little bit above in the chat, if anyone wants to to check that out. But I see earlier, Toyin, you had a question, which actually Tracy might be able to help provide some context on as well, but around tips for transitioning from academia to industry. Toyin, do you want to introduce yourself and expand on that?

Certainly. Hello, my name is Toyin Ola. I am a Behavioral Health Data Analytics Program Coordinator for a state behavioral health agency, and right now I am focused on trying to get a data workflow together for what is essentially a police criminal justice and behavioral health program to divert individuals from the criminal justice system when they're experiencing behavioral health emergencies. And we've learned that really a big component of that will be getting data, 911 call data, from public safety answering points. So to Frank's point, I've spent a lot of time learning about the emergency communications world, so that I do have some sort of shared language with them since I was completely naive to that world beforehand. But I guess, you know, my question is tips for transitioning industries when you're new to an industry, don't understand the language, and even on a more technical side, you know, there are some analyses that are more common in certain industries than others. So any feedback along those lines would be helpful.

Tips for transitioning industries

Yeah, that's a great question, and I think for data scientists, so true, because the data science skills are applicable in so many different contexts, but then you have to get up to speed on that context. I guess maybe I'll speak from my personal experience. I mean, one is just like that confidence and that you can learn it. Like, just, you know, you know, you can learn a lot like, just, you know, there's always going to be that learning, and it sounds like you already have that, and like encouraging that in others, just that sense that, like, no, you're not going to know everything, but what I have confidence in is my ability to learn less.

In terms of, yeah, particular tips is one situation I've seen, because I also worked in high-performance computing, is maybe that the computational person comes in sort of feeling like they have the answers and not necessarily listening or asking the right questions. So coming into it, you know, you sound like you're already doing all of this, so, you know, just like with that perspective of listening, but also thinking of the right questions to ask, so spending some time not only, like, listening, but then formulating questions for people to answer, and sometimes people are better about answering them, like, in a conversation, and some people like to have time with the questions and write them, so sometimes I'll provide different mechanisms for people to give me information or feedback, because people work in different ways, so I've found that to be helpful.

I'm in this similar situation now. I'm six weeks into RStudio, I think, so I'm trying to learn and ask questions, and then I think, yeah, that other strength that it sounds like you're already bringing to the table is that synthesis, so writing down for yourself to synthesize the information or write it down and share it back and say, you know, is this right? I don't know if those are actually helpful.

It is. Thank you. I appreciate it. Thanks, Tracy. Sep, do you have anything else to add to that, too?

I mean, I have a lot of interests, so I'm, like, between a lot of different kind of areas. I think for me, and I've also hired people that are from outside of healthcare, outside of pharma, exactly what Tracy was saying, I think being humble, but also tenacious, where, you know, if it's an area that you want to kind of go explore and apply, you know, again, your data science skills are kind of applicable to anything. It's that domain knowledge that generally people need to show in a particular interview, and so, yeah, if you're tenacious enough to be like, yeah, I could do it and you have that confidence to do it, but also recognize that you don't know everything and there may be nuances and be receptive to listening and getting that feedback. I think those are attributes that are fantastic to have because I love taking people that are outside of that area just to bring a new perspective into it, but they have to also have the right personality to be able to do so in a manner that's not like, oh, like, you know, dismiss or disregard, you know, whatever anybody else is saying because they know it all.

To that end, like, even doing, you know, small projects with data in that given space that you're interested in also goes a long way to show that, oh, yeah, this person's actually pretty serious about, you know, pursuing this, you know, they did this in-depth analysis in whatever sports analytics or something like that, and so pairing those two together, I think you should be totally fine whichever way you want to go. Thank you, I appreciate it.

If anyone else has anything to add there, I see Frank, you asked a question, kind of a follow-up question from a point made earlier around data science being, is it different than data science infrastructure? Frank, would you want to expand a bit on that?

Yeah, okay, so as this whole conversation's been going on, I've been thinking about, and it started when Trevor had mentioned the garbage in, garbage out, and talking about his experience there, but in the supply chain world, I go and visit distribution centers, and right, being in retail, so I work for Target, so being in retail, we have a ton of distribution centers around the country, and I've been in supply chain for quite a few years now, and when you go into a distribution center as you're touring the warehouse and walking around, if you are paid hourly to keep it organized and pick up the trash, that's what you do. If you are a manager walking the warehouse and you see a piece of trash on the floor, you pick it up. If you are the CEO of the company and you see a piece of trash on the floor, you pick it up. Everybody that walks through there is responsible for making sure it's organized and clean because it's safe, and the truth is, if our distribution centers don't operate, we don't have product on the shelves, we don't make money, and we don't profit, right? That's bad for everyone, so everyone feels a responsibility to do that no matter what your level.

I think about the data industry and data-driven companies, and Sep, you had talked about a data engineering group versus a data science group, and once people start to have this perspective where they say, hey, this is my job, this is what I do, and they allow garbage data to flow through their pipelines because they were responsible for the pipeline, not what's flowing through it, I feel like that's a really bad environment and mindset. I see that responsibility for having the right infrastructure, infrastructure being like the data because that's the information. Everyone that touches it, somewhere along the line, should to feel that sense of commitment to make it better, and I guess then my head would, that's kind of where I left off and thought like, man, the data science part, you can't just be responsible for data managing and building algorithms, and then on the other side, like the infrastructure teams, the teams that are building tools can't just be responsible for the tools and say like, hey, tool, go out into the world, like good luck out there. It's got to be a team effort, and I know that as your company gets bigger, like I work at Target, like it gets harder and harder and harder to instill responsibility for that because there's a trade-off, right? We only have, every individual and team only has so much time, so much bandwidth and resources, but it is absolutely critical if you think you are going to become a data-driven company. Tangent over. Does anyone have thoughts?

It's got to be a team effort, and I know that as your company gets bigger, like I work at Target, like it gets harder and harder and harder to instill responsibility for that because there's a trade-off, right? We only have, every individual and team only has so much time, so much bandwidth and resources, but it is absolutely critical if you think you are going to become a data-driven company.

Standing ovation, Frank, standing ovation. That resonates with me 100%. I totally agree, like if you're going to do work, do quality work, but I don't know, at least for me, the reality is that many of the people that I've engaged with are on both sides of the coin. Some people are like, hey, it's just a job, I'm going to do what I'm told, and it's sometimes, especially with data, it's like it takes that little bit extra to be like, am I, just let me just double check, am I really getting what I'm supposed to be getting? And then, and if you don't do that, then it's, again, it's on the next Target employee or whatever to go pick up that trash, right? So I don't have a particular solution. I know in our industry, you know, patients' lives effectively are on the other end of it, and so that's a big motivation for many people. That's why they do what they do. It's not easy, but yeah, it's also not, a lot of it also is domain knowledge. Like if you're not a clinician and you don't know what hematocrit is, right, like how do I even know what I'm looking at, right? So that's the problem we end up having is we have to have dedicated like informaticists that are also looking at the data and you can't rely on a data engineer that might not even know what the medical medical information is, right? But I totally agree. I wish people kind of did that.

I think Hugh has raised his hand. Hey, so no, I was just going to offer a quick aside. There was a really neat presentation I think Colin Gillespie gave last week on practical advice for arm production that talked a little bit about infrastructure versus the data science piece and, you know, it's all of our responsibility, you know, to make sure things work, but at some

point, I really like what he said, at some point role specialization is crucial because you wouldn't ask IT to build your model for you. You wouldn't ask them to, you know, do your exploratory data analysis, all that stuff, but you also wouldn't ask the data scientist to make sure you have a update schedule or that, you know, you're all using the same version of R, that kind of thing. So it's neat you talk about how do you find that balance and depending on the maturity of your program and the, you know, the amount of support you have, you have to find it at your level.

Yeah, great point. Thanks Hugh. I just put that link in the chat as well. I just found a recording there.

I know we're just at the top of the hour here, so if anyone has to jump, no worries. You have to run to another meeting. I do see there's one other question left that we didn't get to yet, so I'd love to chat more about that if Sergio is still on. Would you want to introduce yourself, or you already introduced yourself, but ask that question live. Yeah, sure. Well, it's something I'm living right now because all the companies I'm working with right there, they don't have anything. They're starting from scratch and it's super interesting. It's basically a progressive transition into the data science culture, but eventually I know our objective is to be able to implement like RStudio Connect tools, something like that, but maybe hearing about some experiences, maybe you can share about small companies that just started with these data science tools and when is a good point to start thinking about or start implementing other tools that is going to be part of developing teams, data science teams, for example.

Starting from scratch with data science infrastructure

Anybody else? I mean, I have a very specific experience with this. I don't know if anybody else would like to jump in, but I have a belief that you have to have your data in order before you get your data scientists and analysts. A lot of times it ends up being reversed and then you have a data scientist who may have pretty reasonable experience with engineering and getting the data in shape, but that's not necessarily their area of expertise. And so as long as that's understood that that's going to be technical debt and it can be addressed later, okay, then that may be fine, but generally you're going to want to get your data in order and have a very clear strategy of how you're ingesting that data, how it's going through its various stages until it gets to its resting state to have then products or whatever built on top of it.

So I would say that would be, that should probably come first. And in many cases, it would be, if possible, it'd be great that it's done with someone that has analytics data science expertise as well, because sometimes the data engineer doesn't have that other side of the coin. And so the two together would be the greatest kind of start where you have a data engineer and you have a data scientist analysts, and then however they prefer to choose. I mean, I'm a little bit R biased, but you could also do this in Python and then you kind of build your tool set and architecture around that, right? So once you have your pipeline, you can then say, okay, well, how are we going to do the analysis? How are we going to make this scalable? How do we deliver our products? And you now have kind of at least some kind of foundation for a small team, literally two people, very talented people, but two people at least. So that would be, that's how I would, if I was to do it over, I would do it that way. I've done it probably in two different iterations of that too.

I have a belief that you have to have your data in order before you get your data scientists and analysts.

Thanks. Does anyone else have anything to add to that?

Mike, I feel like your analogies in the chat also apply to this piece as well.

Yeah, thanks. I, I was actually thinking a bit, jumping to a different analogy about cooking, right? You know, everyone can cook at home with the equipment they have to hand. You don't need the very best, but you know, if you want Michelin star cooking, then you've got to have the right people with the right tools, the right skills to bring it all together. And I, I think it's easy it's easy, just what Seth was saying, it's not easy, but it's easier when there's a very small group, because you can have that one-to-one dialogue that says, okay, I know what you're doing. You know what I'm doing. We know the tools that we have. Let's come together and get this done. When you go to a really big organization, and this comes back to something that Frank was saying, there's a tendency for people to grab the tools they know, hack it, whatever they need to do to get today's work done and get it out the door. And the problem there is that in doing that process, they leave a damn mess everywhere. You know, they leave the pots and pans lying about and everything's, you know, messy and they don't clean up afterwards. And so, you know, having, as the organization grows, having people that know what they're doing, you know, know exactly what's expected to get what they need to do, do what they need to do, and then leave the place in a tidy framework, because they know that they're going to have to do it again, you know, in a minute or in an hour or tomorrow. And so there's that kind of chef-y thing of mise en place. You chop up, you get all the pieces nice and tidy so that when the orders start coming in thick and fast, you know you can get your job done.

And my, you know, I work in a big organization and, you know, there's an awful lot of people who are doing their own way, you know, and I kind of like the tidyverse thing of it's opinionated, but it's opinionated for a reason because you can then, you know, reproduce that pipeline again and again and again using the same kind of tools. You know, I have lots of colleagues that learned R 10 years ago and they're all base R and they're like, yeah, but I can do that in base R. It's like, that's great. I'm really happy for you that you can do that in base R. This isn't a competition. This isn't code golf. You know, I'd like us to be able to do it in a way that I can pass my stuff to you, you can pass your stuff to me, and, you know, we know what we're going to get and we can then be part of that, you know, chain of getting stuff out the door, but getting it out the door, looking good, smart, fulfilling the brief, you know, what everyone wants to see on their plate. I'm sorry, I'm all about analogies.

I kind of like the tidyverse thing of it's opinionated, but it's opinionated for a reason because you can then, you know, reproduce that pipeline again and again and again using the same kind of tools.

I love it. You're inspiring me to go get better organized.

I don't think I see any other questions in the chat, but if I've missed any, please feel free to speak up too. Yeah, I have to drop unfortunately, but I appreciate you guys giving me the time to share a little bit of my perspective. I appreciate everybody else who also shared their perspective and asked great questions. Thank you to Rachel and our studio team to putting this together. I look forward to the future ones as well. Awesome. And Sep, quick question. What's the best way for people to get in touch with you if they have other follow-up questions? Google me, I guess. Yeah, LinkedIn is perfectly fine, but I'm pretty much available on like whatever platform, but yeah, LinkedIn is probably easiest. Okay, awesome. Well, thank you so much, Sep, for joining and sharing your insights. I know you have to run, but thank you all for joining as well and same place, same time next week too, if you happen to be free. We'll also put the recordings up on YouTube too, but thank you all. Have a great rest of the day.