image: thumbnail.jpg
Investing in community to drive organizational change | Dooti Roy | Data Science Hangout
We were recently joined by Dooti Roy, Head of Clinical Data Science Capability Management at Boehringer Ingelheim Pharmaceuticals, to chat about organizational change in data science, investing in internal open-source community, transitioning from SAS to R, and motivating long-term learning and adoption of new tools.
In this Hangout, we explore the instrumental role of internal community in driving organizational change. Dooti shared that during the preparation for the transformation process from closed source (like SAS) to R and open source, she was advised by Rachael Dempsey (creator of the DS hangout!) to invest in community and "don't sleep on that." Dooti reported that following this advice proved successful, as the internal community grew from the early adopters and became a safe space where colleagues help each other understand new concepts and overcome implementation challenges. This thriving community, which includes team members from the US, Europe, China, and Japan, serves as a springboard for testing new ideas, tools, and processes, generating global feedback that helps pull the institutional change forward.
Maybe our favorite part, Dootie shared that strategies like naming internal meetings "beginR" (for beginner office hours) help set the right expectation, ensuring people feel comfortable asking questions without feeling afraid of appearing naive.
Resources mentioned in the video and zoom chat:
🔗 Positron Data Science Lab Session Registration → pos.it/dslab
🔗 R/Pharma Conference Registration (Free Event) → https://rinpharma.com/
🔗 Community Building Tips & Tricks Site → https://pos.it/community-building
🔗 Boehringer Ingelheim's Open Source Davinci Modular Framework → https://github.com/Boehringer-Ingelheim/davinci
If you didn’t join live, one great discussion you missed from the zoom chat was about the struggle of finding time to learn new skills. Many attendees agreed that they are sooooo tired after work and that the learning often needs to be a work assignment to make time for it. This resonated with Dooti Roy's observation that "I don't have time" is one of the biggest pain points for colleagues attempting to learn new tools. Do you have the energy after work to do ambitious learning projects?
â–º Subscribe to Our Channel Here: https://bit.ly/2TzgcOu
Follow Us Here:
Website: https://www.posit.co
Hangout: https://pos.it/dsh
LinkedIn: https://www.linkedin.com/company/posit-software
Bluesky: https://bsky.app/profile/posit.co
Thanks for hanging out with us! 💛
Timestamps
00:00 Introduction
02:52 "What did organizational change look like and how did community help?"
06:19 "Do leaders understand the commitment to open source?"
15:25 "Was this your first time promoting open source?"
18:04 "What does data science look like at Boehringer Ingelheim?"
22:55 "How do you motivate long-term learning and development?"
31:05 "What is your costume for Halloween?"
31:55 "What is your team's workflow and technology stack?"
44:50 "How do you motivate data professionals to document their code?"
48:40 "How do you encourage people to keep up with open source developments?"
Transcript#
This transcript was generated automatically and may contain errors.
Hey there, welcome to the Posit Data Science Hangout. I'm Libby Herron, and this is a recording of our weekly community call that happens every Thursday at 12 p.m. U.S. Eastern Time. If you are not joining us live, you miss out on the amazing chat that's going on. So find the link in the description where you can add our call to your calendar and come hang out with the most supportive, friendly, and funny data community you'll ever experience.
I can't wait to see you there. I am super excited to announce our featured leader, Dooti Roy, head of Clinical Data Science Capability Management at Boehringer Ingelheim Pharmaceuticals. That is such a hard one to say.
I'm not blaming you. It happens to the best of us.
We're going to call it Boehringer for the rest of the time.
Dooti, thank you for being here. I would love it if you could give us a little intro about yourself, what you do, but also something you like to do for fun outside of work.
Thank you so much. First of all, I first admit that this is my first ever data science hangout. So I was honestly a bit taken aback by the number of people and thanks so much to the community who joined here. Happy to see so many of you and also willing to really have a chat instead of a monologue. Libby assured me that this is exactly how it flows.
So I'm excited about that. A little bit about me, Libby mentioned my title, it's a handful. What I can do is break it down in simple language. What I do is lead the U.S. clinical data science team in Boehringer world. That means colleagues who are statisticians by training and also statistical programmers by training. This is the team that I lead. And on my second hat, which I'm also fortunate to wear, I am leading the SAS2R, or open source transformation in the company.
So really cool active space to be in and also the reason why Libby and I connected among other colleagues in Posit. So happy to be here. So over to you.
Thanks. OK, you knew Rachel before you knew me. And I remember when I met you, one thing that you were talking about as we were preparing for this Hangout was organizational change in data science right now. This is a huge topic because we are dealing with like outside market forces that are forcing our hand and forcing us to change and examine the tools that we use in our workflows, the way that we do things. And you said that Rachel had told you don't sleep on community, don't sleep on investing in the community. And you said that you were so glad that you did.
So I would love it if we could chat, start just chatting a little bit about what that organizational change looked like for you, what the challenge was and how you invested in a community and how that's changed the process.
Investing in internal community
Thanks. And this has like many part answer. So first and foremost, let's start with the part on Rachel. You're right. And I'm glad you remember. That's also my first meeting that I met her with Rachel. And I knew from the conversation that she came from a background of having kind of been in conversation with several global companies who were kind of beginning to walk towards this are open source and away from SaaS or closed source solution for coding.
And then I remember asking Rachel, I'm like, hey, what is that one thing you would advise for a beginner of the transformation process that you think people would wish the new one year in? And Rachel's answer was invest into building an internal community.
Right. And when I presented this back to my more senior leadership, the global leaders in the organization, one of the pushbacks I had was, well, you know, like we have so much other things to build. However, tools, templates, infrastructure, documentation process, like must we invest in the community? Maybe it can grow gradually. And somehow the conviction with which Rachel talked about it really convinced me.
So I pushed back and I said, no, we're doing this. And I think that has therefore in the next year and a half to come, has become one of the instrumental pieces of why I think this change management part is going so well internally. From starting with the early adopters, these guys build up the community from ground up. And now it has become a safe space where people just come through questions and the community help each other out to understand things that are different from their day to day work and how to implement it and what are the challenges.
And I think that has therefore in the next year and a half to come, has become one of the instrumental pieces of why I think this change management part is going so well internally.
And it's really lovely to see. In addition, what has also happened, this community is kind of a springboard for us testing new ideas. So if we are to bring to them a new feature, a new process, a new tool, we would just go to the community and say, here is go through it over and then feedbacks. And then suddenly you have colleagues from US, from Europe, from China, from Japan. This is the spread of the data science team in Boehringer. Global feedback flows in that this is good. That's bad. Change that. Update that. Improve that. And that's so cool for product development in general. So those are the two that stand out.
The feedback loop. I remember you saying that in the beginning it felt like there was this push from above to push the change. And now that you have a community and the community is strong and healthy, that the community is pulling the change. The community wants the change to happen.
Leadership challenges in the open source transition
And I wanted to talk a little bit about the challenges that you faced with the transition and what the transition is. The transition is closed source to open source. And one of the challenges that you had mentioned to me was a leadership challenge. Like do leaders who are non-technical people, who are not embedded in the open source world, know what they are committing to when they say we are going to be open source? Could you talk a little bit about that?
Thanks. That's also a question very close to my heart. You know, like my a little bit like reflection on my own background. I was trained as a statistician, finished my PhD and then jumped into a very typical clinical trial statistician role and then found my way through the organization. I was really not really into the world of coding details. Right. I could code to save my life. Sure. Right. I mean, we all did our PhDs in using R and that's OK.
But what really that meant in the open source sense, I think it has been a discovery on my part as well. And as I began to discover the world, how connected it is, like how supportive it is. It comes with a commitment of give and take. It talks about reproducibility. And then you begin to realize, I'm not so sure if everybody understands open source in the same way. And then this is something to learn by itself.
So we started having more conversations with our leadership, of course. Right. So I had the intent from the leadership to say we want to move to R slash open source. And it took a while to explain that R slash open source is not really exactly slash topic. One thing is to say I want to move a programming language away, which means you're recreating or creating new tools. But on the other end, open source comes with a whole bunch of other philosophical angles, I would almost argue.
What I think is that once we started explaining these details and introducing what open source communities look like and what can you achieve, how quickly problem solving, for example, are coming together as a community to realize some of the key problem areas and come up with solutions together, which then also get adopted quicker across the board, create this buy in. The leaders begin to realize that it's a huge advantage and a positive thing.
So that's one conversation. And the other aspect of the conversation was we don't want to be just takers. We always want to give back. So this conversation around, oh, we have created a new code base or a new tool or a new template and now make it open source. And people will be like, do we need to invest a resource? And kind of, yeah, because you signed up for it. So it's easy to say we want to move to open source. But if you don't really walk the talk, then that might be a challenge. So I'm happy to report that this conversation goes very well these days. I think I'm having to fight less and less. Yeah, it's all good.
Less and less. I remember you saying open source is tough because it's inherently anti-capitalist. And so you're talking to these business owners and saying, OK, so here's the thing. We're going to give our competitors our tools, right?
Community polling and giving back
I am going to put I've been putting some some polls up in Slido just so that we can ask some questions of the community and one of them was like, have you experienced an open source transition at work from closed source to open source? Less than half said yes, more than half said no. I think that the no side is probably we're already using open source, right?
Another question that I asked is, do you have an internal community at your company? And it's really 50-50. As I see people vote, it'll change a little bit and then it goes back to 50-50. If anybody out there does not have an internal community and you would like to have an internal community, please fill out the survey when you leave and tell Rachel and I that so that we can help you get to that point and give you some resources. We'd be happy to like meet and chat with you about it, too.
I'm going to turn on one more poll here. That is, does your, if your company does utilize open source technology, do they give back? Because I know some will, like Eric Nance is going to say yes, right? Like pharma versus we give back. There's a lot in pharma that do, but I know that I feel like most people in other industries don't. So let's let that poll run and hop over to Slido for questions which are piling up.
Okay, maybe just one small addition, Libby, to asking for Rachel's help in community building, because I'm a big fan of her in this regard. You know, like some of the details of, she shared and Rachel's here, so maybe she can jump in and if she remembers what she told us a year and a half back, one of the things she said is it's very important to think about the psychology of a person coming in to ask you a question and how you name your meetings, for example.
So, you know, a meeting, we have decided to call our internal meetings Big In R, so Beginner Office Hours, but Bigger Big In R. And I think by naming it this way, you immediately translate to your audience that it is completely till to be a beginner, you don't know anything, all questions are good. And I didn't really recognize, like, this little tidbit suggestions and what kind of impact it can therefore have in building community, the community gave feedback on it later on, right? What makes them feel at home and part of it was how the meeting was set up with the very name, setting a different type of expectation that you don't need to be afraid to ask, like, questions they think is naive or stupid, it's all fine.
Just again, a shout out here to really think this through on what you do with your community and also, like, take feedback from the community to understand how they are feeling when they approach the organizers of this community and what kind of space they have created. It's really helpful.
Thank you so much, Judy. I love getting to hear those lessons through you, but I also have to give a shout out to the community at John Deere, because that's where I actually learned that about the beginner office hours. And I remember their team telling me that just by changing the name, I think they had double the amount of people who showed up for that community event.
And I'm just going to share, this is a bit of an older resource now, but if it is helpful for anyone, I had tried to collect a lot of community building tips and tricks from people and put it into this Quarto site here. So, definitely check that out, but also let me know if there's other things that you want to add to that.
And I remember their team telling me that just by changing the name, I think they had double the amount of people who showed up for that community event.
Yeah, and thank you. I mean, with that tidbits of learnings that you shared, and you see how this gets multiplied across, right, when you tell us something, immediately, this has started happening in four countries around the world. And the user, it's a global community, but it's also regional communities with different people alternating and taking, raising their hands and leading that, and then they bring their own flavor to it. So, it's a really, like, community-driven-led piece.
And I'm so happy to see that. It's, like, it takes less and less intervention from leadership side, from my side, to do this, do that. It's, like, they're running their own show. We sometimes have conversations to kind of take into account how it's going, something that we can do to make it even easier. And that's really about it. So, it took, like, one year or roughly so to get it all up and running and in a stable state.
I think that's a great thing to call out is the time horizon, because a lot of people are like, we're going to make a community. It's going to be thriving on day two, right? No, you're looking at, like, 12 to 18 months of time horizon for things to kind of get going and to see the benefit. And during that 12 to 18 months, you are pumping a lot of time and effort into making things good. All right. Well, I want to hop over to the Slido and ask Marlene to unmute and ask her question. Rachel will help Marlene get unmuted.
Because I'd love for her to ask that live.
Marlene, I just noticed that your name wasn't spelled out as Marlene. Sorry. There we go.
It's Anne Hankin. I know it's her email address.
You're on mute. I pressed S to unmute, but maybe it's not letting you. There you go. There she is.
Okay. I keep hitting unmute, and then it mutes me back. Sorry. Hi, Judy. Thanks for joining us today. I don't know what time zone you're in. So, I just hate today. It's been time.
Good question. So, you know, like my background, I mentioned I was a trained statistician. And in between, what I was doing is driving statistical methodology in the company. So, in my world, it was always peer reviewed. It was always open source share, because that's methodology is only thriving as many people adopt it.
And that has been also a growing journey for me to realize how amazing, intensely vast and cool this community is, right? There have been so many aspects that I've discovered on the way, and that has served as sort of like an upscaling opportunity for me. But also, then, to bring some of these notions that I observed as a statistical researcher back into this transformation piece, which means that we always shared code otherwise while writing publications. Why can't we do it?
Well, thank you so much, Marlene, for that great question. And I saw one pop up from Kenneth. Kenneth, I never know how to say your name. It's either Rosever or Rosever. Let me know how you say it. But I'd love you to ask your question live. It's Rosever. Can you hear me? Rosever. Thanks, Kenneth. Yeah, actually, it's Rosever, but we don't even say it right. So don't worry about it.
These are common slogans you will probably hear across many other companies, but Boehringer began to think about what does that actually mean? In order to implement data driven decision making, you need the infrastructure, you need governance, you need availability of tools and templates at your fingertips. You need basically a way to extract insights rather quickly from the raw data until the formulated story that you can tell with it. So all of these kind of open areas to develop.
And then what the company did was focus on building these capabilities and partially 2 years ago, I started leading 1 of these capabilities, which was called analysis and visualization. If you think about it from a. Pharma companies perspective, it's like the entire range of things you could do. Yeah, you do visualize, but you also analyze, you help reporting, you help them decision making therefore.
And I think that's what I would say. The biggest change has been in my observation. Like, we used to have cool implementations before, but they used to be more like, oh, it's a pilot here. A pilot there, the impact was limited, but then this company really started thinking about or enabling us to think about implementation at scale. What does that mean? Right? I don't want a pilot. I want all trials. Clinical trials that are running across the board to take advantage of some of these tools and templates and that takes a lot more than, oh, yeah, have some people who know how to write code.
And this is where also the open source community becomes so valuable because you realize if you want to create your universe from ground up in a silo by yourself, it takes a heck a lot of time kind of makes sense why some of these product would take 10 years to develop. But if it's open source, we have a different opportunity. We can come together as a community, decide that this is a priority across the board and try to create something together.
So I think that's really one of the ways I had explained to my leadership. What's the benefit of leaning into it? Right? Everybody wins kind of in this situation. It's like, you don't have to do things that you don't find useful, but you can also contribute to it. And if they find it useful too, then you have a win-win situation.
Everybody wins kind of in this situation. It's like, you don't have to do things that you don't find useful, but you can also contribute to it. And if they find it useful too, then you have a win-win situation.
Okay, cool. Thank you. Thanks.
So the previous discussion about being inviting to like small group development meetings was really good. Thank you for that. But how do you encourage people to actually want to do in the first place? So we've had multiple iterations. I work at the University of Kansas, and we've had multiple small groups gather over the years. And it's been really great to have time to learn about our Python or Tableau or whatever we were working on at the time. But how do you encourage people to want to continue in learning and development? Long term and thank you for that. I think.
You know, 1 of the undeniable truth about working in a company, or in the industry is that. And then everybody's like, you go ahead and do it, right? It's usually comes with that added burden of explaining to the company and how is learning pottery will help the company's bottom line. And I think this is also the same approach we have taken. We try to really understand. Is there value for the company to do it?
If you are tasked with, oh, here is this, like, 50,000 rows Excel sheet and good luck with that. That's a burden, like, people do it with all the diligence in the world. But they also realize that this is extremely tedious activity, which obviously no pharma company can avoid because it's very important to make sure that the patients are safe while taking this experimental drugs, right?
This is where we could come in and we did. To create a shiny application. Where initially, when we presented, they said, oh, this is really cool. It helps us explore the data. But, you know, what I do, I also need to write down notes next to a patient record.
This is what a safety physicians are meant to do. They write down notes. If they have some findings, they, they don't want to, like, forget about it. So, next time when they come back to the same file, maybe now another 10,000 rows have been added. They don't want to forget the remaining 20,000 that they have already reviewed.
And that was a challenge because shiny by itself was not meant to, uh, you know, include data and information as we go, but we have an internal team of software developers. Um, and I can talk about the team in a bit because I think it was really cool thing that we have it. And these guys basically took it from there and now they have enabled a shiny dashboard possibility, which has data tracking and review.
And now it fits the needs, right? And immediately you see that. Oh, totally. This is now built for my personal use case. I can see the value in it and also makes my life easy. So that's like 1 example of.
Motivating people and just to draw the connection back to our own communities. It has impact now. These people are community are also then involved in different projects and trials. And now they are stakeholders who know of this development are saying, oh, that's very cool. I would like to have it. Can you make it happen?
Create now they have an added incentive to say, oh, I would like to lean into this and see if I can learn it and deliver it to my team. So it's kind of like works in this cyclical fashion, right? But it starts with understanding the needs of the users or the stakeholders and then really driving that point home by delivering the product, which helps.
Community, mandates, and organizational transformation
Yeah, and I wanted to call Renato in Renato if you are available to unmute. Just to chime in with a little bit of his experience. From a community perspective. Yeah, sure thing. Can you hear me?
Well, yeah, so I get hired by the UN to teach macroeconomic data management with our. To central banks or central bank analysts, and I've noticed like I've done this with a few countries, and I noticed that when the director says, oh, this is really great. This is the way we're moving forward. You know, like, drop your Excel files, your random stuff, and this is the way we're moving forward. Then then what I teach sticks, right?
Otherwise, it just kind of fizzles into oblivion. And like I get an email from someone, you know, like, where was that that module that you taught us or something like that in a few years? But but you can see that they still use their own their Excel files and their like old methods.
I think thank you for that. I totally agree. I think we have that also internally. But what I can say is the internal mandate was very abstract and broad in nature, right? So the sentence that we had was we are going to move to R and open source by 2027.
It's cool that it's at a timeline and it has that abstract broadness to it. But then the job of the team was to really interpret what does that even mean? Practically speaking, and then that ballooned into this massive initiative, which spans so many different aspects.
And then that's that's why maybe I kind of mentioned like the organizational transformation. I really didn't mean teaching people how to code in R. It's so much beyond, like, I can give an example.
So we started talking about using interactive dashboards to present data stories to our senior stakeholders, right? These are the guys who are essentially yes or no decision makers, whether to fund a trial and an asset development. Everybody was on board in the first place to say, oh, yeah, this is definitely a better way of doing it. And then came the question, well, is this validated?
Well, guess what? We don't even know what is the definition of a Shiny validated. So these are like practical challenges. And sometimes we just take a step back, innovate on that part first. So what we did was have a work stream started focused on validation. What do we mean, right?
Even if you're creating a single use program, if you're creating an interactive Shiny, what does that mean? So how you define validation? Can you operationalize that definition? Have you updated your processes? Have you communicated to your stakeholders? And that became this whole square as sort of like, I see that as one of the fundamental pillars that any organization need to think through.
Because these are the questions you will have from your stakeholders, right? If your stakeholders and your leadership is used to this idea of validated outputs, it translates to something. And then you cannot just say, oops, we don't care about validation anymore.
So some of the challenges has come, but also very cool opportunities for colleagues to explore and innovate in those spaces, right? Because it's pretty open at the moment.
I like that. I like the focus on that this is an opportunity to innovate. It's not just a top down mandate that you have to like deal with.
Workflow and technology stack
Fantastic. Thank you, Renato, for hopping in with your experience. I loved it. I wanted to ask a couple of questions that can be pretty quick to answer. One is, what's your costume for Halloween? Are you a Halloween person? Somebody asked that. I thought it was really fun.
I don't, I'm afraid I don't think I really celebrate Halloween routinely, but I have been known to like don on fancy crazy hats. And walk around in a wizard's robe. But that was years back in college.
I love it. Mine's usually I have one witch hat and it doesn't fit my head, but I clip it on anyway. That's my costume.
I was not prepared to answer what's my Halloween costume. I always love, there's always a question that ends up in Flutter that's funny.
And I wanted to ask one more that I think will be quick, but it's on topic. And it's just a context building question for us. It was, what is your workflow and what software do you use? And I think that that is a nice thing to do at the little halfway mark or more than halfway mark here.
Maybe the technology stack that your team works with to get their job done. I think it's primarily Jira, right? If you talk about technical product development, that would be the answer. But in terms of creating workflows, because a lot of the work from our, my leadership team's perspective with me is the strategic piece, right? So it's like, there's no given defined workflow already. You literally have to innovate as you go.
There we use Agile tools like business mapping and Kanban, kind of like Klaxons to like really like draw it out piece by piece. We have a tool called Draw.io where we use that to kind of figure out process maps, but it's basically really dependent on what you're doing and what tool fits.
So that kind of answers the workflow part for sure. And we use Jira too. Everybody, let me know what you use. Do you use Jira or do you use Asana or Monday or something else? I'm curious. Erin says Jira.
And then what about tools that are data science tools or development tools that your team uses? So I see what you're asking. So basically it has layers. What we have from Posit is our infrastructure platform, which is called Copernicus internally. What it is, is Posit Workbench, Posit Connect and Posit Package Manager combined into three. And as a wrapper on top of that, we have our own in-house developed statistical computing environment, which we call RISE.
And this environment has been developed with a specific mind to connect to the Posit piece, but also at the same time, be open to whatever other ideas to come later on. At the moment, it accommodates SAS and R with the intention that the SAS part will be off in the next couple of years to come. So that's what it looks like.
I think that that's probably part of what the question was. It was like, we're talking about this transition from closed source to open source, but then we never actually said it's from SAS to R or from SAS to Python or whatever. So thank you, anonymous asker, for asking that. And in the pharma industry, still SAS is a major player, right?
The case for open source: speed and simulation
Many companies still depend on SAS for their solutioning. I think where we started seeing advantages and pushing it, also in my previous role in the stats methods lead, what we observed is more and more the regulatory demands talk about simulating. And if you are running large scale simulation, I think nobody really in their sane mind uses SAS. Sorry, SAS. And then you are using evidently R, but then it wasn't really like as sophisticated solution as was it offers.
So it was not cloud-based. It was an internal HPC servers that we used to use. And that was really lengthy and time consuming. So when you are facing regulatory questions or somehow time limit on, you need to do your simulations, but also you don't have all year to do it. It became very clear that colleagues were struggling internally to somehow like manage the scope of simulation without compromising the science part, but also deliver in time so that you can answer the questions that we're initially hunting for.
And I can tell you one of the first instances of implementing Copernicus, one team reported that their simulation has now become 500 times faster. And those are numbers and metrics that no one can deny. And it speaks to everybody, to the colleagues who are hands-on running this because they feel the pain, but also to the leadership who is now being able to go to their leadership beyond and talk about my team just delivered this like much, much faster than what they would be able to do before. And these things matter.
And those are numbers and metrics that no one can deny. And it speaks to everybody, to the colleagues who are hands-on running this because they feel the pain, but also to the leadership who is now being able to go to their leadership beyond and talk about my team just delivered this like much, much faster than what they would be able to do before.
So creating these kinds of stories, emphasizing those in the community kind of builds that credibility over time to say over and over again, you move, it's not just a language shift. It actually creates value, which you never would be able to tap into using the previous infrastructure and tools. And now you can.
That's amazing. It's so funny that you say that because some of the other people that we've talked to at other pharma companies, actually, and I'm thinking about like Takeda and others had that similar discussion. Harvey Lieberman also mentioned this about speed. It's like there was something that took a really long time and people were either doing it manually or doing it sort of half manually. And it was taking weeks or months to do it. And then, you know, we take the time, we invest the time to get it switched over to an open source solution like our, and then it takes a day or it takes half an hour or it takes somebody, you know, part of their work week instead of several work weeks. That's amazing.
Getting started with open source
I wanted to also shout out a anonymous question that was asked in the Slido that I also turned into a open text question in Slido for everybody to answer. There are so many great responses there. So definitely go add yours. And it was, how can someone start getting into open source, especially if they have been working on proprietary software all their life?
There are several examples there. Judy, did you want to add anything really quickly? Absolutely. And I think that's a really good question and it's a real question, right? It's not just like I want to, but how do I do that?
I think long story short, my way and maybe here, I think Marlene was asking about some of the failures or struggles that have happened. I can kind of relate to that. This is one of my previous initiatives and how it went. So by definition, by design, I was in my career developing new approaches and trying to convince people that this is the way to go and move away from the old stuff, like forget about open source in general, in stats methods.
And oftentimes I have struggled with that, right? And one of the biggest pain points that I heard over and over again was I don't have time. I'm already so full with my plate that if you think that I can carve out 20% of my time and learn something, forget it. It's not going to happen.
And one of the, I mean, it's just so true, right? We struggle with it. It's easy for leaders to say, oh, we'll just learn it. How? Where is that extra amount of time going to come from? And that's a struggle.
What I decided to do differently in this particular topic of learning R or other tools beyond SAS was try to overlap that with their day-to-day, which basically means you are anyway tasked in executing a particular activity and you used to do it in SAS. Let me find you a buddy or a mentor or this community organizers who are our coaches.
If we can see you can implement the same template using a different language, but now you are serving your own work at the same time serving the community. And this type of overlapping model worked very nicely, is working very nicely for Boringa. And I'm talking from a global company perspective. There's so many hundreds of people.
It's not always easy to find this, but I can tell you that I've been able to find this in many cases. It's like pushing back on ideas where you hear from people, oh, I can just retrospectively implement something and immediately realize that retrospective implementation is basically just add onto your plate without you being able to tell the company and here is how I served business too.
So it takes a little bit of courage and boldness to say, I'm confident I can start on this as a real deliverable. And if I face problems, my community is there to back me up and solve those problems. So it's not like you don't feel alone. We are with you, but at the same time have a little bit of courage to come into something, which then is a very nice topic for the leadership to say, and here is how I also did my work.
So it takes a little bit of courage and boldness to say, I'm confident I can start on this as a real deliverable. And if I face problems, my community is there to back me up and solve those problems.
Building Shiny apps in clinical trials
This model has been working nicely. I can tell you an example of one of the things that any clinical trialist have to do is monitor data as they come in all throughout the conduct. And we do this in a quality review meeting, which happens every few months.
And in the middle of this year, we wrote an email to the entire company and said, and we are going to do this in an interactive, shiny fashion, period. Let's move away from this thousands pages of PDF outputs, which makes everybody's life difficult.
And there was this pushback to say, how are we going to implement these things on top of the daily work? What we were able to do is take people who were already in charge of implementing these reviews, instead of creating SAS based PDFs, they were implementing Shiny with the help of the community support. And it simultaneously helped build up templates for other later trials to use.
So it was basically like you learn suddenly your work can create value, not just for now, but also become scalable so that it can be used for the future work that you may have or your colleagues may have.
Modular code and quality control
And I made this reference to the software development team, which is I really need to plug them a little bit here. One of the struggles we have had, and many of you have experienced when it comes to developing tools and templates, many of us think of this as a one-time activity. Like you just do something, it works for the moment, but we forget that there is a part around maintenance.
So creating a template is easy, maintaining it and making it still working after two years is the hard part, right? And one of the pieces that we also realize, and I think many of you will acknowledge it, it is so hard to work with someone else's code, right? This is not an easy activity to say, oh, give me some code and I will make it. Fantastic, it's not easy.
So one of the things we decided to do was help the community by building modules or smaller code chunks. And then the community's job is to take those modules and then they can mix and match in whichever way they sound fit and build a bigger app out of it. And the quality of modules is controlled because it's built by the same team who are at the top of their game in software development and writing a good piece of software, they know how to do that. So that was one way of a solution to scale this piece, simultaneously controlling for the quality and maintenance with.
I love it. They're giving their colleagues Legos to build and put together things that work and they're high quality Legos. So everyone might put them together differently, but at least the Legos work. I like that. And I saw lots of heads nodding when you said, it's so hard to use other people's code.
And Tim said, it's hard for me to work with my code since me in the past was not kind to present day me and commenting my code. Tim, yes, absolutely. I will come back next day and go, what was I thinking? What was I doing? And now in today's day and age, what was chat GPT thinking? What was Claude thinking? Why didn't I write down or discern better what they were doing?
Arsenis, I saw that you had a question in Slido. Would you like to unmute and ask that live? We've got about 10 minutes left with Judy. Sure. Thanks Libby. And thank you very much Judy for meeting, for talking with us today.
So my question is about this thing that Libby just mentioned, which is documentation. So you actually mentioned when you were introducing kind of what you do and the change that you all are making at Bargainer, that documentation is one of the things that you kind of work on. And what I'd like to know is how it is that you all motivate data professionals, especially folks that are beginners in open source and all that who typically might not care
that much about documentation when you get into a project to document. What are some of the things that you do for that?
That's a great question. And I can tell you, it's not just the initial people getting in, nobody likes documentation. Let's just like admit it.
It's also good to admit that we are in this highly regulated industry workspace. Documentation is something you do as part of your job requirement. This is not an optional story that you say, oh yeah, I did something I forgot. And now there is no traceability or reproducibility of what you have done. And therefore what decisions you have made, this comes up in audits.
And then you get like a smack down from your leadership, say how bad this is not good. This has happened historically. I mean, my company is also more than 150 years old. They have learned their lessons. Therefore there's emphasis on maintaining documentation.
What was critical for us to realize that because we are defining a whole new world, there was no templates on documents that we could rely on, right? Some of this documentation had to be created as we go. I touched upon this topic around validation. That was one example, right? You're writing these documents to define the process and who does what and who is checking. You also have to talk about where is this saved. And there's this layers to the story.
What I think internally, I can share that it has been not so strugglesome because people already had the habit of documenting things. It's just about saying, and this process comes with this documentation responsibility. So it wasn't like, this part was not movers and shakers for colleagues because they're already used to that conception of like, yeah, if I do something, I need to document.
I hope that helped. Yeah, documentation's always rough. I know, right? But here, LLMs have given us an advantage for sure. I think one of the first things people started talking about LLMs where can we essentially take away some of these tasks and do it in a more efficient way. And I think that's definitely, I mean, you have probably seen examples in the industry that this is already happening. Some of this typical documentation that I created over and over again, enough templates to feed it to the LLM and say, okay, I need this and that. Can you create a draft? Great. And then you get the tedious task of editing it and make sure it's right.
Motivating awareness of open source developments
Arsanis, thank you for that question. Let's hop over to a question that I saw Eric asked. I hope Eric is still here. Yes, he's still here. I like this question, Eric, because I think that Doody has something to say about this one.
Yeah, I'm really enjoying this content so far. Sorry, I was fashionably late. When you get in the shining of the bug and things happen, time flies, but nonetheless, we've all, like I said, in life sciences, I'm part of life sciences too. And we know that open source is growing immensely into how we're developing these innovative capabilities. It's one thing to consume, like the open source content out there, such as using farmer's packages and whatnot. But it's another to help motivate people in your organization to want to keep up with the pulse of what's happening, even if they're not necessarily contributing back to it, at least being aware of what's happening. So I have some thoughts, but I'd like to hear what you have to say about how do you motivate people on your team or your colleagues to not just consume these tools, really see what's happening out there and really being abreast of what's happening in the latest and greatest in some of these developments out there. And thanks.
I think that question really resonates with me. Libby was right. It's basically this challenge also, which comes with having worked in a closed source environment where we would just be concerned with our templates, our way of doing it, our way of reporting it, and then forget about what are other people doing.
And I think it has started to permeate. I would not say we are already there doing 100% of external awareness, but what has helped is my team is essentially a group of connected individuals who are no longer just working in one silo that they are responsible for. I see this like mashing effect to say, like coming together to help each other and they all see the bigger picture. Part of that is then sending these individuals or encouraging them to attend conferences like Fawzit conference, for example, or a fuse or a DIA. In the US, I was hosting a couple of sessions in the regulatory industry stats workshop, like going to places, engaging with colleagues, and then also bringing that back to simulate that excitement to say a lot of things are happening and there is a lot to learn and a lot to give and a lot to like exchange on.
But I think that this is going to take some time to really like permeate at every level of the organization. I would still say this is particularly noticeable in people who are directly involved in implementing pieces in open source and internally in the working group, but I would not say that this happens over and across. This is something, a work in progress and also an important piece.
I've been telling people this part around giving back. I think some of you may know that we had made this entire modular framework, which we call the Vinci open source. It's available. You can check it out, but it's also thinking about all these resources that we are creating internally. Somebody talked about learning. How do we start internally? We have created some really cool resources, like people are handwriting resources sometimes or coming together from different sources. My first thought was, can we make that open source? So hopefully next year, we will also push some of these internal developed tools and templates and products in the open source space. Each of this should help the community in small and large way to get there together.
I'm curious, like your own thoughts. You said you had some thoughts of your own. I'm very curious to also.
Well, yeah, you capture a lot of my thoughts too. There was a situation that really motivated it. I had a really talented colleague come to me with a prototype Shiny app that was doing like high level exploratory data analysis. And I said, this looks good, but have you heard of Teal? And I showed him this and literally he's like, if I had known this, I would never have built this. Teal does everything I need. So it's just like trying to figure out how to put these little breadcrumbs in and obviously user group, internal user groups are one way of doing that. So that people come at one kind of cadence, regular cadence to inform them. But yeah, like you said, a well-connected team and just trying to spread that knowledge out, but in a way that the other people want to join on that learning journey and not just feel like they're forced to. I would agree with you, it's a work in progress.
It's about awareness, right? If people aren't aware, they can't be excited about something. And Judy had said, with the health of the community getting better inside of Boehringer, that you've even had people who are talking to you about jobs and those people are like, I don't really care what the payer of the job is. Do I get to use the new tools? Do I get to do the fun, cool stuff? So you're obviously doing something right with creating a sense of anticipation around this cool new stuff.
It's about awareness, right? If people aren't aware, they can't be excited about something.
Yeah, I love it. I was so glad to hear that and what Libby just shared as an anecdote, it actually happened internally. So I was talking to a colleague about a certain role and usually would expect people talk about titles, compensation, benefits, even team. They ask who is this time? I just want to use all the latest technology. Don't make me work the old way. Yeah, I was like, whoa, that's inspiring. I had never heard somebody talk about it. So there are these really positive signs, I must say.
So we are all feeling community-wise very hopeful that this is going to be a good way to get into and be successful. And I wish all of you guys the same. I think we are at one hour mark, Libby.
We are, I know. We've gone straight to the line. Well, Judy, I really, really appreciate you hanging out with us. I'm sure you've got some place to be. I want to thank you for all of the wisdom. Thank you to the community for hanging out with us. You can save the chat if it has been helpful. Top right corner, there's three dots. And we are really excited to see you next week. We have Arielle Dror, Director of Data and Analytics at Bay FC. That's a team in the US National Women's Soccer League. So I'm really excited about that. She's going to talk all about working out loud, which is one of my favorite topics. Join us next week. I will see you in the Discord server, everybody, between now and then, and on Blue Sky and LinkedIn. Judy, thank you so much for hanging out with us. Thank you so much, guys. It was really, really a pleasure. Thanks a lot. Go find Judy on LinkedIn and say hello. Goodbye, everybody. Have a wonderful day.