Transcript#

This transcript was generated automatically and may contain errors.

Hey there, welcome to the Paws at 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.

Olivia, I'm so happy to have you here today. Could you introduce yourself and tell us something that you like to do for fun? Sure. Thanks, Libby. Olivia Hebner. I'm a data science manager here at Summit. We are a data science and analytics firm working mostly with federal clients. We have some state clients and commercial work as well. Something fun. I say that this is something fun half the time and something I hate half the time, but unfortunately, I'm a runner. So I like to torture myself by running half marathons, full marathons. It's enjoyable sometimes, but not all the time. And yeah, as Libby said, when I'm nervous speaking, I kind of tend to turn it into my own stand-up comedy routine. So throw me some pity laughs. It'll make me feel better.

So this is Olivia's talk. And it was called Same Data, Different Tools, Visualizing with R and Python. And it was one of those things where you were like, here is a thing. This is another thing that looks exactly the same, like visualization. Here are some maps. It's like, this one was made in R. This one was made in Python. Can you tell the difference? I couldn't tell the difference. I teach both R and Python. I use both. I advocate for both. Often, we see people advocating for one over the other or fighting about it. So seeing a talk where you're like, look, we're all doing the same thing. They're different tools, but we can do the same stuff was so refreshing.

Background and career path

So I'd love to hear a little bit about your background and how you came to coding and analytics. Sure. So I was an econ major in undergrad and then went econ for master's as well. I just had great professors who really made me interested in the subject and then joined Summit as an analyst. So I've actually been here for almost nine years and have climbed up the ladder in that time. I think it's interesting, right? Because when I was in undergrad, data science was not an available major. So econ to data science kind of made sense then. But from there at Summit, it was a lot of Stata coding at first, which then turned into some R coding, which then turned into Python and circling or cycling through different tools. So Stata was the entrance.

Stata was my first love is what I always say. I will take no Stata slander. I understand that it's not as good as R and Python. I admit that. But you can't get over your first love.

Transitioning into management

Well, another thing that you have grown into is managing a team of data scientists, but also software engineers. How did you decide that you were ready to make that leap between I am an IC, I'm an analyst, and I'm now ready to manage people? Is that something you were excited to do and grow into? Did you have any hesitation there?

Yes, I definitely. Yeah, it's a funny question. I had hesitation. If you ask my manager, who I think is on this call, I definitely went to him at some point and said, I don't want to be a manager. This is not the next move for me. He was like, OK, I hear you, but let's give it some time. Let's have you attend some goal settings with me and some year-end reviews and just get a little taste of what it's like. Over time, I really grew to value spending that time with my staff and his staff and thinking about not just how I can grow my career, but how I can help grow other people's careers and what that can teach me. I learned so much from everyone around me. I don't know if it was one moment where I woke up and was like, I'm going to be a manager this year. I'm ready, but a slow evolution of being open to the possibility instead of just saying adamantly, no, this is not for me. I think my biggest holdup there was I was afraid that I would lose my coding skills or not be able to code as much.

I don't code as much on some days and then other days I'm coding from nine to five. It's about making that balance and finding what works for you.

I don't know if it was one moment where I woke up and was like, I'm going to be a manager this year. I'm ready, but a slow evolution of being open to the possibility instead of just saying adamantly, no, this is not for me.

I think it can be scary to do things that you're unfamiliar with. So presenting the opportunity to do it in baby steps and just see what it's like really helped.

Managing when your team knows more than you

The next one is anonymous. It says some people on my team are more knowledgeable than I am in certain areas, and I worry that I shouldn't be managing them. How do you lead effectively in this situation? I relate so much to that question. I genuinely think everyone I manage is smarter than I am. And I don't know if that's imposter syndrome or if that's just what it is, right? I mean, you can't be the smartest in the room about every single thing and data science encompasses so much. So I think, you know, I look at it as I'm never going to like, one of our staff is an expert in R Shiny. I mean, she can spin up a fabulous app in an hour. It would take me probably two days, right? I'd get there, but it would take me longer. And just looking at it as, okay, how can I help best guide her? How can I think of like, all right, you're great at doing this in R Shiny. Why don't we try this widget? Like, why don't we try expanding your skills in X, Y, or Z? But I'm never, you know, I think if you start comparing yourself to your team, you're going to feel understandably a little bit down about yourself. But also like, that's amazing that you have a team that, you know, feels smarter than you in subjects. And like, I think that really shows that you're a great leader and you're supporting their careers, right?

The work at Summit Consulting

Sure. Yeah. I'd say it's very diverse. We are very well known for federal loans and working with federal credit programs. So think of things like small business loans or USDA gives out farm loans or we have international loans going out from DFC. So lots of loan data, which is how much is the loan? What is the interest rate on the loan? What is the risk associated with the loan? There's kind of a, I don't want to say a standard structure, but right, you generally need X, Y, and Z when you issue a loan and that data is consistent across data sets. So it makes moving from one federal loan program to another a little bit easier, although they all have their unique complexities.

But then we also work with the department of justice. And so we help on litigation analytics, supporting testifying experts and running the analysis for them. And that data can be absolutely anything, it's whatever DOJ is investigating at that time. So currently we are working with some dispensing data from across the country and looking at prescriptions that were filled. And then we also on the non-federal side, we work with a great company who is known for education analytics. And we work with them on the New York state education data. So looking at, I believe it is middle school data, looking at grades and trying to predict attendance and performance and are those correlated. So it's kind of a little bit of everything, which makes it really interesting for me.

Tools are, we're a little bit scattered everywhere. Part of it is client preference. So a lot of federal clients prefer to work in SAS. So we do have contracts that use SAS and that's their preference. They're our clients. Great. We'll use SAS. R, Python, I would say that we're kind of phasing out of Stata. We also use SQL pretty heavily on our projects. We're in the cloud a lot, AWS and Azure, you know, git of course everybody's favorite.

Yes, we do. Or Shiny, I guess too, right? As dashboards. Yes, we definitely have quite a few R Shiny apps. Most of them again are tied to those federal loan programs. So we're going in and users can input loan information and it'll spit out. Okay, great. Here's what your cash flow looks like. So we have three or four R Shiny apps that do that currently. And that year I just remember being very big on Quarto. There were a lot of Quarto talks. And so one of our staff, when we got back, we were working on a report for the New York state education department. And she was like, let's do it in Quarto this year. Let's, you know, let's try it out. Let's see what it is. So we shifted from, I think that was a shift from R Markdown to Quarto that year. But yeah, we, everyone always wants to try out the new thing and, you know, we, we see how we can make that work on projects.

Privacy and security with federal data

So I work with this little academic organization that studies politicians using surveys. And so I'm always keyed into privacy and I'm curious about how privacy works with some of the data that you're using. So like, are there just rigid protocols that you follow? What was it like to learn to use those?

Yeah. I mean, that's such a great question and I think it's like ever evolving, right. And maybe it gets a little bit more scrutinized every year, even. I'd say obviously, as a company, if we are hosting the data ourselves, we have to meet certain standards, lots of ATUs, which are authority to use and ATOs, authority to operate that government agencies issue to us. And it shows that we've passed, it's genuinely hundreds of security tests that we have to show we're monitoring and improving upon and, you know, reach X, Y, and Z metrics every year. A lot of that is helped if you're hosting the data in AWS or Azure, which are both already FedRAMP compliant. FedRAMP is a big security that puts out, you know, policies and rules and regulations around specifically government data. And so, you know, if you're building a new product or a solution, if you're trying to sell the government something and you get it FedRAMP approved, that goes a long way in getting rid of security issues.

When you're working with like your federal clients, for instance, how do you go about like, if you're redoing their legacy systems or code or whatever, getting buy in to rebuild that? Because I had a lot of issues with this at the Postal Service, like getting buy in to do that and also like not offending people that built these systems because sometimes that causes the whole thing to like, shut down.

Yeah, no, absolutely. I hear where you're coming from. We also one of the things that Summit is known for is model validations where we go in and review someone's model before it goes to audit. And like we are on their side. We want to make sure they pass the audit. But people can be very particular about their model and they don't want to admit that it could change in X, Y or Z. I think a lot of it and this is maybe not a helpful answer is having that like trust and relationship with your client. So I in no way can take credit for our latest modernization effort where we moved everything from Stata and VBA to Python and an Azure front end UI. But that is a client we've worked with for over a decade. And the leader on that project has been here at Summit for a decade. So they've had that relationship. And instead of focusing on what's wrong with the current model, they really focused on why this new modernized solution would be better. And it also was given to them in small small bits. So instead of completely overhaul your model, get rid of all of Excel and Stata and transition to this technology that you don't understand, you will not be able to code review it yourself. It was, you know, well, we could really reduce model runtime because right now it's taking two hours to run this model and we could make it run in five minutes. And like small, small bits and maybe even small proof points if you're able. But I do really think a lot of it is that relationship building, which does take time and it can be really frustrating when you're like, I know this would be better.

Giving a conference talk

Sure. Um, yeah, we did a small data challenge at Summit through, oh my gosh, I should have brushed up on this before, but Data Viz Society, I believe, every February they do a Black History Month Data Viz Challenge centered around W. E. B. Du Bois and his artwork. So it was recreating something that was drawn in, I believe it was pencil, in R or Python, and you want to get it as close to the original as possible, right? You don't, you want to try and make it feel like it was hand drawn instead of like, this is a graph that I clearly created in ggplot. So we did a little challenge, part of it was to have Python users learn R and vice versa. And then we were like, the idea was, oh, we'll have everyone vote at the end and see which output is better. And then at the end of the challenge, we got to our two outputs and they look identical. So the vote was basically 50-50. No one could decide, oh, that's Python, that's R. We all just agreed that they looked the same. So it was a really interesting find. I actually really expected there to be some sort of difference between the two.

And then giving a Posit conf talk, I hate public speaking, just like absolutely hate it. And it's something that I've been working on my whole career. I've terrified before. A fun fact is that I started college as a theater major. So that clearly didn't work with stage fright and being afraid of public speaking. But I went to Posit in 2024, thought it was amazing, was so envious of all the speakers, and said, you know what, you can do this, like submit a talk, you can do it, push yourself to do it. And the data challenge was kind of just like a natural, this could be a really cool talk to give.

Preparing for public speaking

Is there anything you do to prepare before giving a talk since they make you nervous? Oh, yes. Practice a million times. Like practice more than you think you need to practice. I was pacing my hotel room for two days straight just saying the slides over and over. And I know that sounds boring. Like that is honestly really boring advice. And it's honestly also kind of painful in the moment because you're nervous and you'd rather be doing anything else than focusing on this thing that's making you nervous. But you know, the more you do it, the more it's in your brain, just the less likely that you're going to anything is going to happen that you're not expecting it to happen. And I'd even say to like I practiced in front of my coworkers. Right. I took their feedback. I practiced with everyone staring at me silently. I did it virtually, you know, different environments and just again, practice, practice, practice.

Yeah. Practice, practice, practice. I before my conf talk, I spent a week cat sitting and dog sitting for my friend who was going out of town. And so every single day, twice a day when I went over, I gave my conf talk to her animals in her house. I love that. So I had this like a captive audience that couldn't tell me anything bad. And they were very cute and fuzzy. And that worked so, so well, because I tell you what, I gave the talk to my husband and I hated that experience. Zero out of 10. Don't recommend. He told me all kinds of things I didn't want to hear. But what was really helpful is if you sign up to give a talk at Posit conference, you're going to be put into a pool of people who are prepping with coaches who are professional speaking coaches. So you're going to be practicing over and over again already in front of this group of people who are all invested in you succeeding and getting really good feedback from a professional.

And then giving a Posit conf talk, I hate public speaking, just like absolutely hate it. And it's something that I've been working on my whole career. But I went to Posit in 2024, thought it was amazing, was so envious of all the speakers, and said, you know what, you can do this, like submit a talk, you can do it, push yourself to do it.

Motivating diverse teams

We have an anonymous question that says, I manage a team where some people focus on doing their job reliably while others are really high achievers. I respect both mindsets, but how do you keep both groups motivated and engaged? Yeah, absolutely. I think again, like it's hard, right? Because every human being is different. We're motivated by different things. I have found a common ground between, I'm going to just say all tech people, data scientists, software engineers, you name it, is challenging them in some sort of way, or maybe not even challenging, but giving them an opportunity to learn something new, to do something a little bit different. That has worked out really well for my team and other teams here as well. So, right, we did that data viz challenge that I presented on. We just actually did another data viz challenge across the entire firm where we encouraged data scientists to team up with staff who had expertise in federal credit loan programs so they could create a very specific federal credit visualization. Pushing them to say, okay, do it in Python. You're great at R. Try it in Python. Giving them little opportunities that break up the monotony of their day-to-day. I found has worked really well to reset their mind and reset their motivation and just get them excited about something.

Making sure every voice is heard

In my team, some members reflect quietly while others jump straight into discussions. Any tips for making sure everyone's voice is heard? Yeah. I mean, believe it or not, I actually started my career as a quiet person and no one at my company believes me, except the ones who were, you know, here during that time. But all of our new staff are like, you, you are the loudest person at this company. There's no way that you ever started quiet. So, I can, like, approach it from being that person as well. And I think it's about building comfortability and building trust within the room and within the team. I do call on people, but I call on people that I know, you know, gently, and I lead it with a question that I know that they can answer, that I know they have an opinion on. And again, I really think it's grounded in that trust and, you know, feeling comfortable to have them say, I don't know, like, I'd love to look into it, but I just don't know. And like, always having them realize that's an acceptable answer as well. But I do think that it's really important to give, you know, shy people the chance to speak and trying to make them feel comfortable the best you can. And sometimes that's just, like, being vulnerable yourself, right? And saying, you know, I don't know what my opinion is on this. I'd love to hear everyone else's thoughts.

Career advice

Is there a piece of career advice that has really helped you or that you like to give other people or maybe something you wish you had been told or learned much earlier in your career? Yeah, I think I hinted at it a little bit already. And I will be honest, I heard it on my very first data science hangout ever. And I've tried to think of something better. But I just think it's the best and it's do it scared. Right? Like if you see someone and you're like, wow, I wish I was giving that talk, like submit a talk, even though it's terrifying. It is terrifying. I agree. But just, you know, do it anyway. Try it anyway. You're just as skilled as the people everyone else is. You're just as skilled. You're just as capable. And the only thing holding you back is nerve. So just do it. And then if you hate it, you never have to do it again.

And I've tried to think of something better. But I just think it's the best and it's do it scared. And the only thing holding you back is nerve. So just do it. And then if you hate it, you never have to do it again.

Like in the beginning, when somebody asked the question about like, hey, I think that everybody on my team is smarter than I am. How do I manage them? There are so many people in the chat who were just saying like, yes, that's how everybody feels. That's how we all feel. However, you're feeling you're not alone in feeling that way. And also, you'll probably surprise yourself.

One thing that I we have an internal sort of little mini Posit internal conference where we get together for work week and that's coming up in March. And I will be giving a lightning talk there about my experience with meeting people to have on the hangout where everybody that I talk to has this hesitation where they're like, I don't know why you want me to talk at the hangout. Like, why me? I don't have anything important to say. I don't have anything interesting to say. And sometimes the people who are most adamant about that, about the fact that they have absolutely nothing to share that anybody would care about, turn out to have so many questions asked in Slido that we can't answer them all, right? Like, they are so interesting to other people. And they had no idea. You really truly don't know until you go talk to people. But I guarantee you, you're more interesting than you think you are.