Shannon Pileggi @ The Prostate Cancer Clinical Trials Consortium | Data Science Hangout

Transcript#

This transcript was generated automatically and may contain errors.

Hi, everybody. Welcome back to the Data Science Hangout. I'm Rachel Dempsey. I lead Customer Marketing at Posit. Posit, if you are joining a Hangout without knowing who we are, we are the open source data science company building tools for the individual team and enterprise. Thanks so much for taking the time out of your day to hang out with us. The Hangout is our open space to hear what's going on in the world of data across different industries and to connect with others who are facing similar things as you. And we get together here every Thursday at the same time, same place.

We do have one week we're taking off August 15th for the week of Posit Conf. And so I know I've asked this many times before, but if anyone from the Hangout will be there in person, please let me know in the chat. I'm trying to get some people together one of the days of the conference as well. But if you want to join us virtually, you can as well. And let me, bad at multitasking here, let me put this into the chat. If you want to register to join the conference virtually too. And I always like to add, if you're watching this Hangout as a recording and you want to join us live in the future, we'd love to have you join us. Just make sure it adds it for 12 Eastern time. I know people really enjoy connecting with other attendees here in the Hangout in the chat. So if you want to use this time now to introduce yourself, say hello, maybe share where you're based or something you like to do for fun, feel free.

I love to see people connecting there and making friends from the Hangout. We're all dedicated to keeping this the friendly and welcoming space that you all have made it and love hearing from you no matter your years of experience, titles, the languages you work in, or the industry that you work in. Another quick note, if you're hiring, you can also share those roles in the chat too. But today, if you want to jump in and ask questions or provide your own perspective, you can raise your hand on Zoom. I'll call on you to jump in. You could put questions in the Zoom chat and feel free to put a little asterisk or star next to it if you want me to read it. And then lastly, we do have a Slido link where you could ask a question anonymously too.

But thank you all again for joining us. I'm so excited to be joined by my co-host today, Shannon Pileggi, Lead Data Scientist at the Prostate Cancer Clinical Trials Consortium. And Shannon, I'd love to have you kick us off with introducing yourself, share a little bit about your role today, but also something you do outside of work too.

Sure. So my name is Shannon Pileggi. My background is in biostatistics. I was a college professor for a few years, and I worked in market research for a few years. And more recently, I've been in clinical trials for the past three years, and it's my happy place now. Outside of data science, I recently got into a gym in my local town. It's a community-owned gym. I go six days a week. I'm really into weightlifting. And I was thinking about it this morning, and I feel like it's not entirely unlike what we do in data science, and that you're constantly pushing yourself to failure and iterating and figuring out what works and moving on and getting stronger.

It is your first job, and it's not going to be your forever job, right? Like, your career is about iteration. You're going to take what you get in your first job, and hopefully it's a good one. But you're going to learn what you like to do, and you're going to learn what you don't like to do, and you're going to iterate and move on from there.

Building inclusive communities

I see Neil asked a question a little bit earlier that was, could you give any leadership advice on how to bring people together and get a diverse range of views and ensure people are heard?

Bringing people together and getting a diverse range of views is really about creating a very safe place for people to talk, like a place like this. So, you need a safe environment, and that comes in a lot of different ways. There's quite a bit of infrastructure and thought that goes into creating that safe environment. So, for example, with R-Ladies, some things that we have are our code of conduct that explains, like, what sorts of behaviors, like, will not be tolerated in our events. And other things that you can do to provide that infrastructure or provide suggestions or ways that people can communicate in that environment, which Rachel does so good at here, right? She tells you, you can drop questions in the chat, you can post questions anonymously, you can star them if you want to read them yourself, right? It's really explicit how to interact in this environment. And the person on the other end of the call, like, doesn't have to guess about, like, what should I do? Am I going to go on camera? Like, what's the process here, right? There's no guesswork for that person. So, that, like, tension and trepidation that a person might feel about interacting in that environment is removed because you know exactly what to do. So, I think when you are trying to build diverse communities, you need to set up that infrastructure to make people safe, but also to guide people on how to interact.

Switching sectors and open source in pharma

Hi, Shannon, I'm also a fellow co-organizer of R-Ladies, so just the Amsterdam chapter. It's a bit later in the day for me. But I was actually interested in asking you a bit about something you said much earlier on about how you're now on clinical trials and that's your happy place and you were a professor for a while and then you were in market research. And I'm just kind of wondering how did you, one, decide to switch between those very different sectors and then also how you navigated it?

Yeah, I mean, when I left teaching and it was like two factor decision, one that I had to relocate to be closer to my family, and I actually like really loved the department. I was teaching and I feel like the best place in the world is like an undergraduate statistics program that truly cares deeply about teaching and teaching well. And I had so much fun there. But I also explained earlier that I was a little frustrated that I wasn't doing my own data science problems. I wasn't advancing my own data science knowledge. I felt like I was further behind than my students in programming and R.

And so I decided when it was time to move, if I couldn't teach in the best place that I thought was the best place in the world, I was not going to teach anymore. And so and I needed to advance my own skill set. And so the first position I took in market research was like a really nice fit for me to learn more about industry and how things operate outside of the academic world. It was different, you know, but it wasn't quite my favorite flavor of science. And I realized I kind of missed that more rigorous environment. And that's so that's why I'm so glad I ended up in clinical trials because it blends what like and pushing the edge of what you can do on data science, but also has that like nice, rigorous environment to it as well.

So then I have also recently I saw that at the Posit Conf that there's going to be this summit about open source and pharmaceutical companies. And I always thought that those two may not go hand in hand. So I am curious about sort of how you use open source tools like R and R packages and things like that in a environment that needs to be very sort of closed and secure and sort of how that those two are balanced.

Yeah. I mean, I'm super lucky that the team I came into like it's a brand new team and didn't exist before, which means that there was like no legacy code existing like to protect. So we were a brand new team. We were writing code from scratch, which meant like we could choose our own tool set and our tool set of choice was R and that was clear, you know, that that's the direction that we were going. As far as like industry standards, like it is accepted now that you can do R programming in clinical trials. But that's not the pushback or the barrier as much — the pushback or the barrier is that like a large portion of the industry is trained in SAS and a large portion of the pharmaceutical companies have a ton of existing framework built upon SAS macros. So that is like really hard inertia to change both the training and the technical infrastructure. But you'll see that a lot of larger pharmaceutical companies have their own entire software engineering teams now where their jobs are to build R packages for internal processes to make that technical transition easier. And so a lot of the pharmaceutical companies are moving in that direction towards R or at least trying to move the needle.

Moving into a managerial role and code review

Shannon, I know I think it was your blog post last year where you did like a year summary and you mentioned moving into more of a managerial role. Can you tell us about some of the lessons learned in moving into that managerial role that you share with some new managers here?

So, I was more at first like an individual contributor, just writing my own code, and then our team grew, and then I had one direct report, and it grew again, and I had two direct reports, and it grew again, and I had three direct reports. And then all of a sudden, I was in a place where I found that I might have been giving the same feedback individually to the three direct reports instead of having a cohesive place where I documented my expectations and made those clear upfront. So, that was something that I had to move towards to make sure, and it probably wasn't a something I could have done in the beginning because I didn't know what my expectations were. And specifically, this is around code review and code submission.

So, part of my managerial role is that the new contributors will submit code and submit pull requests, and I'm looking for two different things in that code. I'm first looking like does the code actually do what it needs to do, which is one question. But sometimes part of the harder question is did the code think about the data correctly? And in that sense, what I mean is that when you work with clinical trials data, like our data comes to us in a relational database type format with like 90 different data sets. And so, there's a lot of nuance in understanding how the data capture works, and that takes experience to get used to. So, it's not just like did your code work right, but like did you bring in the appropriate data from the appropriate data sets to make this conclusion?

And so, I'm reviewing a lot of code, and I was giving feedback individually, and then I realized I'm giving the same feedback over and over again, and I've got to document this to make it clear. And I think once I did that, people were like oh my gosh, you told me, and now I understand. Like now that I've seen it all written out, like this makes sense. Thank you.

Yeah, I think code review is like such an interesting thing, because there's like this very traditional context of code review in software engineering, where you're like writing functions for packages, you're doing unit tests, you're making sure the code does exactly what it says it's going to do, but that's like not the type of code review that I do. Like we're doing code review for data analysis, and I think that's a much different flavor. I actually made my thoughts on it public on GitHub. If one of my friends on the call right now wants to go find that link and drop it in, that would be great. And you're welcome to read it, look through my thoughts, contribute anything back. And I made it public because I think it's — some of the things are kind of like still kind of internal to our team, but some of the things I think can more broadly apply to other teams. And I put it in a GitHub repo and not a blog post because I feel like it's going to be changing and evolving as I continue to think about it, and I also would love like other people's feedback on it.

No, I mean, I think when you're working in a team on code, you kind of have to have, especially if you're working long-term, or like, I don't know if this is a one-off project or not, but like you kind of have to hone in on like what your preferred style of coding is so that we can all read the code independently and smoothly without having to guess what's going on, kind of have things in the same order and lining up.

But in terms of like tips for code review, maybe I can work on that more. I mean, one of my goals over the next year is going to be for me to do less code review and for my direct reports to be doing more code review of each other and me so that they flip the role, and I think as we do that conversation, we'll learn more about like processes about what works for everyone in terms of it. But it's a pretty painstaking process, to be honest, to go through everything line by line and validate the results as well to make sure that the tables are coming out right, that the numbers are lining up, but it has to be done.

My initial tip, because you mentioned it, would then be allocate a lot of time, because it takes more time than you think, and probably as much or I'm not even sure what fraction, but possibly as much time as it took that person to write the code.

Community participation and career growth

Libby, I see you asked a question a bit earlier around the R community. Do you want to jump in?

Sure. I lost it, but it was definitely like how has being active in the community changed your trajectory, do you think, or opened up opportunities for you? I'm often talking to people who are very, very hesitant to start interacting in the community or start posting things or just start talking about what they know, and I wish that I could give them an example of an experience like, look, this is how expansive it could be for you.

It is so expansive. I would say there's immense — I mean, you put in the investment when you do start interacting with the community, and it's hard and it's scary, but the returns on it are so immense, right?

Previously, before I started blogging, or maybe even initially when I started blogging, I felt like I had to figure everything out by myself, and I would just spend hours googling things and being kind of unhappy and chasing dead ends and not getting the answer to my questions and being frustrated. And then when you're part of a community, you can just ask for help and skip that eight hours, and you can get help pretty fast. So there's huge time savings there, and it's fun, and you get to learn what other people's areas of expertise are.

And then when you're part of a community, you can just ask for help and skip that eight hours, and you can get help pretty fast. So there's huge time savings there, and it's fun, and you get to learn what other people's areas of expertise are.