Data Science Hangout | Alice Walsh, Pathos | Improving an Interview Experience

Welcome to the Data Science Hangout. If you're joining for the first time, I'm Rachel and it's great to meet you. If this is your first hangout, this is an open space for the whole data science community to connect and chat about data science leadership, questions you're facing and what's going on in the world of data science.

Also on the Data Science Hangout site, so you can always go back and rewatch or find helpful resources shared there too. We always want to create spaces where everybody can participate and we can hear from everyone. So there's three ways that you can ask questions today and jump in. You can raise your hand on the Zoom chat, so I can call on you. You can put questions in the Zoom chat and feel free to just put a little star next to your question if you want me to read it out loud instead, but we also have a Slido link that Tyler or Hannah will share in just a moment in the chat, so you can ask questions anonymous or both of you at the same time.

But with all of that, I'm so excited to be joined by my co-host for today, Alice Walsh. Alice is VP of Translational Research at Pathos. And Alice, I'd love to just have you maybe introduce yourself and a little bit about the work that you do to get us going.

Yeah, absolutely. And let me know if you have any problems hearing me. I had a yogurt-related microphone incident before this started, so I'm hoping that everything is going to be good. Yeah, so thank you so much, Rachel, for having me. As mentioned, I'm Alice. I live here outside of Philadelphia, where I work in drug development, and I think there's probably been, I don't know, I'd be curious to hear on this call how many people do pharma or research-related activities, because it's an interesting area, because pharma development has always been very data-rich, right?

There's always been biostats teams who are involved in analyzing clinical trials and all that sort of stuff. And what I do is I'm involved much earlier in the process, so I've worked with everything from drug discovery teams who do target identification, like what should we even make a drug for, all the way through to drugs that are in registrational trials, like big phase three studies, and we're trying to understand more around what are the patients that are responding to drugs and benefiting from drugs versus those patients who aren't.

So I would describe myself as training in computational biology. So I have a couple degrees in engineering, bioengineering, biological engineering, and I got into drug development because I was really interested in working in patient data. So I did a PhD where we did some really interesting projects where at University of Pennsylvania, you could work with like a neurosurgeon and get access to like brain cancer samples. And I thought like, oh, wow, this is so amazing, I can look at like 20 patients. And then I interviewed with people in the pharmaceutical industry, and they were like, oh, we have like 1000 patients. I was like, oh, my God, that's amazing. I want to do that. I want to work with patient data.

And I want to try to understand, you know, what does disease mean at a molecular level? And how could we use that information to make smarter drugs, or to identify the patients that really should take our drugs. So I can say a little bit about where I've worked and things I've worked at Janssen, I worked at Bristol Myers Squibb. So those are two of like the largest pharmaceutical companies, where I was doing like biomarker type work leading teams who are working on those different aspects of drug development. And so then most recently, now I work at a startup company. So it's like the complete opposite of working at a giant company where you have lots of infrastructure and IT and things like that.

So I moved over to oncology therapeutic startup where I was employee number two. And so now we have like five people, we're going to 10 people. So it's really exciting to be part of a company where we're starting from scratch.

Moving from big pharma to a startup

What's been the biggest change in moving from a big company to this startup? Other than the infrastructure and all that? I have a lot fewer meetings. I talked to people before about like, oh, what's it like day to day in a big company? And I'm like, well, if you like having 13 meetings a day, it's fantastic. Which I'm a people person. I like talking to people. So I actually didn't really mind having like nine to five, here's a meeting here, here's a meeting there. It's like a very dynamic environment.

But it is difficult if you really need to like put your head down and do some a lot of like critical thinking. And so I was doing all my thinking time in my car, you know, like driving between the office and the house. And now I work from home and I don't have that time, but I just have a lot more non meeting time to like do some thinking, strategizing, come up with crazy ideas, all that sort of stuff. So that's the biggest difference, fewer meetings, because when there's only four people at a company, it's really hard to talk to them all day. By definition, the math doesn't work to have like 13 meetings a day.

Remote work and brainstorming

I wasn't actually going to wait to ask this question of it, because you just said a lot of your thinking happened in driving to work. And I switched to remote work. I know something you are interested in is optimizing remote work and remote teams. So curious to just hear a bit about your thoughts on that.

Yeah, now I have like all the good ideas come like on the treadmill, on the bike, in the shower. Like, I don't have the car time. But yeah, I would love to talk to people about how they're doing remote work well, especially for data teams, right? Like, if I think about the types of interactions that I'm having, like, you don't necessarily want to like make a big slide deck to explain something to a colleague or a coworker. So like communicating effectively when all you have is like Slack huddles and like Zoom meetings and all those sorts of things is a challenge.

Before I joined this startup, I was remote working as well to some extent, right? So thanks to the pandemic, but even without the pandemic, I was on a team where I had direct reports in Boston, New Jersey, and San Francisco. So we were already like a distributed team and we were working with people in Spain. So we were working across like a, I don't know how many hours time zone, which is like an interesting challenge. But I would love to hear how other people are optimizing for that because I just feel like there's got to be a better way. Like people have got to learn something in the last two years of pandemic about how to do remote work really, really well.

I can share about what I just posted, which is about the book Rest by Alex Ping. So Rest is, if you haven't read it, you should totally go read it. It's fantastic. It gives lots of different accounts of different people. Lots of them are scientists. Some of them are technology people. And it's accounts kind of all throughout history of famous scientists who use their downtime to do their background brain work, right? Like the walking around on campus or the mountain climbing or the whatever that they did was really essential to the background brain processes that they needed to have ideas and think and all of that stuff.

I've been working remotely since 2015 and I feel like I was not very good at it for a long time. But I had a director at one point who was like, I need you to like take 15 minute walks during your day. It's like bill them as hours. They're billable hours. This is part of you working for this client. I need you to go take walks. And that really helped like building that into your schedule so that you have the white space to let your brain do background work is really important. Otherwise, you just sit there and you're at your screen for longer than a couple of hours and your brain stops doing all that background process work because you're forcing it to do too much stuff that's in front of your face.

I would like to get any advice on how we can work to make brainstorming work better virtually because it seems like that is an area where people are so excited to get out of meetings when they're virtual. If anyone has any ideas about how to improve on that process, that would be great.

The yeah, the brainstorming thing is the one that I've been trying out a couple of things like there are online platforms like Mural and probably like I've used a couple other ones where you can like drag around stickies and you can like everybody can kind of collaborate on like this big whiteboard together. I found that those can be like difficult to use, but some of the concepts are useful. So like in my team, we've done brainstorming or we just use Google Slides and basically like to prepare for the brainstorming, I do some prep work around like here are the like high level like what if questions that we're trying to get ideas around and then like have people go off and do their own brainstorming on separate slides.

So like it's really low tech like I'll be like, here's your slide. Here's your slide. Here's your slide and everybody write down like as many ideas as they can and it's like the no bad ideas kind of concept and then after like 10 or 15 minutes of everybody doing individually and not looking at each other's ideas. Then we come back together and we try to move them around like categorize them. So like we say like, okay, it looks like there's actually like five ideas here. And so let's start a new slide for each of those ideas and everybody can like move their ideas around. So you can see how some tools where you can like really like drag and drop stuff would be great. But Google Slides is a very low tech way to get the same effect.

I think that's actually maybe one of the places where like everybody calling in is kind of like a great equalizer. Because I know I have colleagues and things who aren't going to like stand up and say like, let me take over the whiteboard and tell you all my great ideas. They might sit back in a meeting. But when you give everybody their own slide, people are really happy to like quietly like fill out their own ideas and put things there. It kind of like levels the playing field.

I think that's actually maybe one of the places where like everybody calling in is kind of like a great equalizer. Because I know I have colleagues and things who aren't going to like stand up and say like, let me take over the whiteboard and tell you all my great ideas. They might sit back in a meeting. But when you give everybody their own slide, people are really happy to like quietly like fill out their own ideas and put things there. It kind of like levels the playing field.

Tool stack and SAS to R transition

Right now we are pretty easy, right? Like we have a GitHub organization account. We do lots of stuff just like locally and are in our RStudio or like Python with VS Code. Everything goes on GitHub. When we need a cloud computer, we use Google Cloud, spin up a VM, use Docker, store our images on the container registry. It's pretty simple because there's like three people working together, which is like a huge luxury to be able to do things simply like that.

But yeah, you know, like I guess I think it's kind of fun being in a small company where it's like if I need to deploy like a Shiny app, like I did this the other day, I was like, OK, I guess I have to figure out how to do that on Google Cloud because there's nobody in the company who's like maintaining an RStudio Connect or something. Right. I'm like, but I can figure it out. You know, can Google around, deploy that, figure out how to get it to work internally and things like that.

I have not. So I've never worked in like a more traditional like biostats in pharma team where they were doing that. I have worked with teams who were like we were like an R team and there's like a SAS team and we want to work together. And I do think that that's challenging, right, to have those kinds of collaborations across. And so you kind of have to meet people where they are and like, you know, okay, let's do some stuff in SAS, let's do some stuff in R, let's find a place where we can collaborate together on the code.

I work for a company where we're all primarily epidemiologists or come from some kind of a research background. And I work in med tech. But basically we were a team that was primarily SAS users. And then we switched over to R. And it was a real, like it was definitely a challenge. We made the change because of financial reasons pretty much. Honestly, I think people are really happy we made the switch. I think what helped a lot was the R for Data Science book. So I really recommend people read that. And I think when a lot of people who use SAS are coming from this research background and a lot of data science, I think, is really structured in this more computer science framework. And so focusing on the tidyverse and going through R for Data Science and really breaking things down into kind of step-by-step research really helped my team get acclimated to R.

Yeah, I feel like that's always a difficult topic maybe around like how much does a team standardize like the tools they have them, people work with, right? I worked in a team where it was like, this person uses Python for this. I use R for that. And we all kind of like did whatever, which has, I mean, I feel like that has some advantages. It's nice. People can use whatever tool they want and feel very much that they can like pick the right tool for the job. When I was at BMS, we basically, everything was in R. It was like, we're just going to use R. It helps a lot for collaboration. So everybody, like Alex was just saying, like everybody's learning together. It actually is like building a lot of collaboration. Everybody can share their code. Everybody can share their experience. So I also really enjoyed that.

Getting started in data science

I would say maybe, though, my number one thing to say would be find them something that you're excited about. Right? So the quote I really like is like, you know, the prerequisite for doing exciting work is being excited about it. Right? So if you're really excited about cooking or like baseball or cancer research or like research in biology, you know, if you're looking for a job, you're going to find something that you're really excited about.

So if you're really excited about cooking or cancer research or like marketing, whatever it is, like find something like that that you can do a project with. Some of the like intro learning stuff, you still need to do it. But if you have a goal in mind, like, yeah, I really want to analyze my data or something, that'll make it so much better to learn and get into it. And then you have an interesting like case study or project that you can also use to like demonstrate that, you know, what's up.

I actually found Twitter to be really helpful for learning about data science stuff. I found like so many resources on there for our. And to be honest, it was kind of overwhelming. So looking back, like, well, I think it was a great place to find things. I think it's also a good idea to like maybe like think about what you're most interested in with doing with data science and just focus on it one at one thing at a time. Like maybe you really just want to do like visualizations. And I think then the tidy, maybe you could do the tidy Tuesday visualizations and you can just focus on that. And I think you'll be surprised how much you can learn just from focusing on one little thing, because even those little things have extra things that come with it.

I also agree that there's like a positive like mental reinforcement when you have like some small successes. Right? Like there's that like joy of like, I made a plot. Like I made a thing and I can like show it to people. I did something. And that really kind of like motivates you to keep going. And I think it helps if it's something that's not like, oh, look, I plotted, you know, like whatever cars, like disc versus horsepower. Like, who gives a shit? Sorry. Like that's not exciting to most people. But if it's your own data and you can say, like, look, I plotted my like steps per day or whatever, it's a really exciting like, I don't know, like what's the right endorphin or whatever that you get from having that, that you can then build on.

Non-data skills and what makes a great data scientist

So I asked if you have any non-data skills or hobbies that you think make you a better data scientist. And I've been thinking about this a lot because I feel like the data science community is really vast and varied. We all have different things we're good at. A lot of us are musicians.

But bigger, like I think there's the question of like what are the soft skills that make like somebody like a good data scientist versus like a great data scientist? And I think that's a really good question. I think we can, people ask like, okay, like what are the skills you need for data science? And I'm like, yeah, like the table stakes are you have to come and you have to know some like programming, maybe something about a database, maybe some statistics, and maybe something about like the business you're in. So in my case, having a strong biologic knowledge of, you know, cancer pathways and things is really important.

But then there's everything else that makes you like go from good to great. So I tend to think like there's the ability to communicate and build your network to influence people, right? That's super important. I think more though, it's also about creativity. So how do you address problems? And then how do you find the right problem? That's a huge thing. I think that's kind of the difference between being like somebody who does something, right? Somebody who does something versus a scientist is that you're looking for the correct question. So somebody might come to you and say like, here, we're going to do this. So like make me do this analysis. You know, give me a P value. But you can come back and say like, actually, I think what we really need is X, Y, Z, this other thing. And I think that's what differentiates people who are just kind of good at their job versus great at their job.

But I could talk about it forever because how you develop those skills, I think are sometimes really not from your day job. They're from like the other things that you go out and develop about yourself. You might not actually cultivate those skills in your nine to five.

Leadership qualities

I mean, I tend to think of good leaders that I've had in the past and what made a good leader. The number one thing is don't be a jerk. There are plenty of good leaders who are complete a-holes, and there are plenty of people who are successful who aren't. So there's got to be a choice there. And so I think there are great leaders who really care about the people that they work with and try to do right by people.

I think we've all had managers where in the past they've made your life miserable, but I fundamentally don't believe that people wake up in the morning and are like, I'm going to come to work and do a crappy job and I'm going to make everybody unhappy. It's all these other circumstances that make somebody be toxic in a work environment, but yeah, so leadership is number one, don't be a jerk. I think the other thing is really communication. If you're a manager, you have to communicate a lot with your reports, but you also need to communicate a lot with your management and then your peers. And especially to advocate for your team.

I think the leaders that I really respect are the people who are all constantly advocating for their team. I go to a lot of conferences and things, and the people who are constantly giving credit for the people who did the work, those are the kind of, I'm like, wow, they're great. I like them. Great job. When somebody doesn't have any acknowledgments or anything in their slides, I'm like, what? What's going on? Nobody wants to work for them.

So, yeah, people who are generous, who communicate a lot, those are kind of the leadership qualities that I tend to hone in on, like am I doing a good enough job communicating? I think the third thing is seeing the big picture. Sometimes people get promoted into leadership positions because they're really good at their technical work. But I think being able to see a bigger picture of how your work fits into that larger picture is very, very important.

Yeah. And you don't have to be constantly giving TED Talks and things about your vision and how you can see everything about the future. But being able to help connect what everybody's doing to the mission of the company or to some sort of bigger impact is really important because it's just very demotivating if you're working on something and you can't see how it relates to something bigger than your individual work.

Communicating data insights to stakeholders

I do think that sometimes the most effective communications I've had have been in person. I tend to not think that. I'm more of a fan of the executive summary where I'm like, here's exactly what I mean. I've seen a lot of people come up with slides or presentations or something where it's like, here's a bunch of things I did. I did this. Here's a plot. I did this. Here's a table. I did this. Here's a table. I prefer if somebody comes to the table or comes to the meeting or whatever and says, here's actually what I think you should take away from it. Yes, here's all the receipts. I have the plots. I have the methods. I have all that stuff, but I'm trying to convince you that this actually means this.

Maybe my biggest communication tip would be have a lot of TLDRs like executive summary like this is the number one thing you should take away from what I'm telling you and then here's a ton of backup. If you want to go through and see the nitty gritty detail, if you want to see my code, if you want to see everything, it's there, but the number one thing is this is what I want you to, I'm trying to persuade you of this point.

Maybe my biggest communication tip would be have a lot of TLDRs like executive summary like this is the number one thing you should take away from what I'm telling you and then here's a ton of backup. If you want to go through and see the nitty gritty detail, if you want to see my code, if you want to see everything, it's there, but the number one thing is this is what I want you to, I'm trying to persuade you of this point.

Yeah, I think I tend to mentor a lot of analysts, more junior analysts at my company and one thing that I notice is you can tell that someone's growing in their career when they go from this is the road I took and then finally telling you why you should care about all the stuff I just told you before versus lead with the headline and start with what people care about and sometimes the people that you're working with will want to know how you got there and sometimes they won't. Then you just give yourself and them a whole lot of headache by doing it that way.

Improving the interview experience

I am. I think that is also another difference between a big company and a small company is that at a big company, you kind of have a lot of baggage, maybe, that people know about the company. That can be a huge incentive because they're like, oh, great, look at this great company. It's a big company. I want to work there. You get a lot of the work done for you of explaining the company. When I've been hiring for my new job, at least until recently, we didn't have a website. We had three employees. I had to work a lot within my network to be like, okay, let me explain to you.

I would actually love, I don't know if we have time yet for people to talk about if they've had any particularly positive or negative remote hiring experiences because that's been something I've been kind of experimenting with is how do I make an interview a good experience for a candidate because at the end of the day, I'm trying to sell them. I'm like, come work with us. I want you to be excited about coming to this job. I don't want the interview to be too long, but I also don't want it to be too short because then they're not going to know enough about the job to come and do it. I would love to hear if people have tips or particularly good or even a particular nightmare. I would love to hear a hiring nightmare situation so that I can never do that.

This was years ago, about 10 years ago, but we hired someone this was before I got into the way I'm using R today and Python and whatnot, but we were giving Excel technical interviews and we were telling all the candidates in advance we're going to be doing Excel technical interviews and this one individual blew it out of the park. It turns out that they must have really, really studied for the basics that we were looking for.

Since then, I have had way more success not giving that kind of technical interview and literally the technical interview is more of a discussion. I would much rather prefer hiring someone if I asked them like, hey, you know how to develop Shiny apps. I'd much rather someone tell me I've never developed a Shiny app in my life, but I use R Markdown every day or that tells me a lot about their ability to actually jump in and learn something new and just their transparency. So I feel like the types of technical interviews I've had a lot of success with the last few years has been more like almost more like a discussion technical interview where I'm not even asking them to whiteboard anything or it's just talking.

No, I love that and that's what I'm trying to get at. If you have any good questions that you ask in those technical interview discussions or something, I would love to hear them because that's what I'm working on. I have this evolving list of interview questions and how we ask them to try to get at some of these things because in some ways I'm much more interested in the process, right? How do people approach a problem and solve challenges that they encounter versus the specific project they worked on? Because they're not going to work on that project ever again with me. It's going to be a new project, so they're going to have to learn something anyway.

I've had like the last few roles I've placed have been exclusively from R related things, whether it's meetups or R hackathons or it's people I've met to the point where I'm starting to feel like I need to make more of an effort to get into a Python related hackathon or a Python conference or something where I might be over-indexing on R folks. But seeing these people at hackathons and the way they just problem solve in a day or four hours of seeing how someone thinks, you really get a lot of insight about these individuals.

Yeah, my company actually does. We do do a technical interview and we give candidates a data set while they're sort of on-site, whether that's them actually being on-site or just being in a Teams meeting. And then we give them maybe like an hour, hour and a half to actually just see what sorts of insights you find with a few very specifically directed questions. And what we're often looking for is not someone to have perfect answers to those questions. Kind of like what you said. It's really about understanding how did you look at the data set? What other information did you want? What do you wish you had more time to do? So you really get to see, like Javier said, how people are thinking, how they work through something.

Yeah, so it went horrifically, but they gave me the job again. The bit which was absolutely horrible and I still get cold sweats thinking about it. I just sort of froze in the interview. It became like a pub quiz on R stuff and that is categorically my worst nightmare of it. What packages are in the tidyverse? At the time I didn't use tidyverse. I was base R and so I ended up going on site and learning tidyverse really well and loving it.

I feel like one that sticks in my head from an interview I did was they were like, please describe in detail the differences between Python 3 and Python 2. I know that they both exist, right? Is that an answer?

Have you ever answered anything with, this is something I would Google? Because I feel like that's a valid answer sometimes. I don't know this, but I know where to find it. So I feel really confident in my ability to Google this. I feel like it's a valid response for me. It's a skill. Honestly, if I ask somebody a question and they said that, they were like, you know what, this is something that I know. I know where I could find the answer to that. I would be like, perfect answer. That's a perfect answer to me. Not knowing, but knowing where to find the information.

Yes. I have been doing that a little bit. I have an optional take-home, which is that here's a data set. Take an hour and tell me something. Use whatever tools you want. Use Excel, use R, use Python, and Abacus. The key thing I want to see is I want to see some written output of what you did and what you would do if you had more time. I'm still kind of on the fence though because I know that a lot of people are so anti having any kind of technical component that I'm still trying to feel out, okay, net, is it better to have it or better not to have it? Is it really helping us make the best hiring decisions?

And Alice, while people are thinking of last questions or if they had any, what's the best way for people to stay in touch with you? Oh, yeah. I think probably LinkedIn. I'm on LinkedIn. I'm also on Twitter. My handle is science Alice, so you can follow me on Twitter for the five times a year I tweet something. Yeah, but like LinkedIn or Twitter are great. Yeah, and I would love to hear people like, honestly, I would love to hear people's interview ideas, especially for like remote hiring, great things that they're doing that they think have worked really well or places that they're finding are really great for hiring. Would love it if you reach out to me on that sort of thing.

I just shared that. I think it would be helpful for everyone too, but thank you so much, Alice. This was awesome. Yeah, thank you. I mean, I had a lot of fun. Really appreciate you sharing your experience with us and thank you for all the great questions too. I'll see everybody next week, hopefully. We will be joined by Travis Girk from the Prostate Clinical Research Trials. Awesome. Thanks, everyone. Have a great rest of the day.