Taking the Lead: Women in Data Science

Data runs the world — and MANY of the solutions KBR delivers for its clients, including the work our experts do to advance defense and national security interests. We’re thrilled to feature one of those experts on the latest episode of the podcast. Kim Cates, senior data scientist with KBR’s Defense Systems Engineering business, spoke with us about her journey in data science, how her cutting-edge work is helping solve customer problems, and how KBR’s culture enables her to take charge of her career.

Transcript

IN ORBIT: A KBR PODCAST

Season 3, Episode 15

Taking the Lead: Women in Data Science

INTRODUCTION

John Arnold

Hello! I’m John, and THIS is In Orbit.

Welcome, welcome, everyone to the podcast. We’re thrilled you’re listening in and keeping us in your orbit.

I’d like to invite you to take a moment and think about how much the world has changed over the past 20 years, particularly in terms of technological advancement.

I’m old enough to remember the novelty of families owning a home PC. Today, at any given moment, human beings hold more computing power in the palm of their hand than sent the first astronauts to the moon.

The world is more connected, more digitized, and at the heart of this period of evolution — this information age — is, of course, data.

At least in the developing world, there’s hardly an area of human life not affected by data and the ways we gather and analyze it and the subsequent ways it’s, hopefully, used to improve processes and make smarter decision.

Naturally, KBR uses data in MANY different ways to help our clients — from facilitating business transformation, to real-time monitoring of critical equipment and assets, to developing tools to help conserve natural resources and fight climate change.

The list goes on and on, including the work our experts do with the United States government to deliver scientific, engineering and technical solutions that advance defense and national security interests.

Well, we’re excited to have one of those experts with us on the podcast today.

Kim Cates is a senior data scientist with KBR’s Defense Systems Engineering business unit — part of KBR Government Solutions U.S.

And she’s here to talk to us about the fascinating and important work that she does, as well as her personal experience working in the wild world of data science.

Welcome to the podcast, Kim!

Kim Cates

Thank you, John. Happy to be here.

John Arnold

So glad you are with us. Well, as is our custom, before we start talking about speeds and feeds and solutions and things, we want to get to know you a little bit. So would you please tell us about yourself and about your journey into the field of data science?

Kim Cates

Yeah, absolutely. So this question actually comes up a lot with me and people I encounter. So it's not a very typical route that I took. I initially wanted to go into cognitive neuroscience. I got into my graduate program. I always really loved statistics and I just learned research wasn't really for me. So I actually ended up learning Python during grad school. I already knew R, which is another language. I did a small project in machine learning. And then my final graduate project was in machine learning, actually building a CNN or an image detection model to predict cognitive impairment using MRI images. And from there I got my first job, or I guess more so internship at KBR. And then once I was done my graduate program, KBR offered me a full-time position.

John Arnold

That's awesome. So you are one of the budding data scientists that have come up through the ranks of KBR internships.

Kim Cates

Yeah, yeah, exactly.

John Arnold

That's awesome. It's always wonderful and promising to hear about the success of those internship programs across the world, whether it's in the US or the UK or where. So that's a really inspiring story. So in reading about you, as is my custom personally, I do a little bit of stalking, internet stalking of guests, and so you earned a master's degree from Seton Hall in Experimental Psychology. I'm also just fascinated by your interest in cognitive neuroscience and then how that translates to what you do today. But anyway, you earned your master's in experimental psychology, data visualization and analysis. So that is a fascinating confluence of subject matter and interesting to think about how all of those areas relate to each other. So would you tell us more about those interests and how it informs the work that you do today?

Kim Cates

Absolutely. I mean, naturally, experimental psychology deals with data visualization typically. I mean, you're going to be visualizing data obviously. And then in addition to the statistics, you're going to be evaluating the data. So any kind of research involves probability testing. So that's pretty straightforward. I think most people would know that. But in my current role, actually, there's a lot that actually translates into data science, machine learning, what have you. I think really you just need to have a good intuition with the data at a very high level.

The ability to do research, oftentimes a customer will come to you or me as a data scientist with a problem saying they want to model this data a certain way. So you have to have the ability to independently research the problem and know what kind of models you can use and look at what kind of papers have done something similar. So that's where the research aspect comes into it.

From a statistics perspective, a lot of the concepts and statistics actually translate over into machine learning, especially when it comes to feature selection and also tuning the parameters, so understanding which parameters to select for the models. So you're typically going to be evaluating, "Okay, given these features, how well does the model perform in contrast to other features or a combination of parameters."

And then a lot of the math. So I mean, machine learning, I would say, and when you get into neural networks also, really involves obviously statistics, and then pre-calc, calc, and then linear algebra. So that's kind of stuff that I also learned in grad school and also continuously learn on the job. I think it really just comes down to the ability to believe that you can learn for yourself. There's a lot of resources out there. There's a lot of YouTube videos, a lot of blogs that you can watch, and it'll explain it conceptually, mathematically. Whatever you want, it's out there, so you just got to look for it and have the confidence that you can do it.

John Arnold:

That's awesome. The term data science, of course, covers a lot of ground. At its most basic, it's extracting knowledge from all kinds of sources to solve problems or improve processes. And you've already alluded to this a little bit, but would you tell us about your specific role and how your work helps solve those client problems?

Kim Cates

Yeah, absolutely. So there's a few different roles I play. So the first thing that comes to mind is I am currently a product owner on a contract that provides NLP for the JSF program. What that really involves is providing technical guidance to some of the data scientists, really making sure that they're not getting too much into the weeds of what they're developing. I think it's very common for any developer, whether it's data scientists, computer scientists, what have you, if you're deep in the code, you really need somebody to look at the bigger picture, making sure you're checking it on multiple steps of the way. So for example, that could be anything from making sure your code runs, making sure it's spitting out the proper results, making sure that you're visualizing everything correctly and making sure it's user-friendly. So all that. So just really ensuring that process from a customer side.

And then other projects that I'm on as more so a data scientist. So I support AFRL. So one of the projects that's of high interest to me is building a model to predict cognitive performance with exposure to GForce. So that's one that's heavily-

John Arnold

Interesting.

Kim Cates

Yeah, yeah. So I've kind of pivoted into doing more so human performance research as that relates to my background with cognitive neuroscience and working with medical data. And then another project is building out visuals for knowledge graphs. In other words, not necessarily knowledge graph, but a meta knowledge graph. So a knowledge graph is a way to represent the taxonomy and oncology of information in a conceptual way. So one of the projects we're building out these meta graphs of medical data, we've done this a few times in my role, is to create the visual for that. And then lastly, the third project... Or I guess that's the fourth. The fourth project that I'm on is essentially recreating a process that I've done. Actually, I published a paper on with some their colleagues to automatically cleanse some maintenance data for aircraft. So yeah, those are all the projects. Two of them are more so NLP and two of them are more so focused on human performance or medical data.

John Arnold

That's so interesting that it covers, I mean, even just in the work that you're doing from asset management and maintenance to human performance.

Kim Cates

Yeah, yeah. So it varies. All those projects are pretty interesting, very stimulating, definitely not redundant at all.

John Arnold

Well, in addition to keeping you interested in your work, what's been one of the more rewarding breakthroughs or discoveries that you've had so far while working at KBR?

Kim Cates

I would say in terms of the work that I do, I think knowing that we can develop a three-step process to automatically cleanse data, that's been really rewarding because that will save a lot of these maintainers a lot of time. In addition to that, I would say building a model to predict cognitive performance given the exposure to GForce. So with that project, it's been a lot of learning signal processing and learning how to align all these physiological sensors together, and then taking these cognitive measures and extracting what we define as a metric of cognitive performance, and then working that all into one overall model. So far appears to be pretty fruitful. We haven't even brand this model on all the data yet, so that's pretty exciting to me.

I would imagine it extends outside of just the work, but I think also kind of mentoring other younger developers and data scientists to really push them to challenge themselves has also been rewarding to me as well.

John Arnold

Oh, that's fantastic. Well, it sounds like in your field, if there is a challenge, you reset the model and go again. So maybe there aren't huge hurdles that you have to cross, but I'm wondering what are some of the bigger challenges that you face in your work?

Kim Cates

Yeah, absolutely. So the first thing that comes to mind is I would say anything dealing with infrastructure. A lot of times the data pipeline in which we're accessing the data to pre-processing the data to loading it into a BI tool or making it accessible to the customer is typically pretty outdated and it takes a lot of work to really communicate that to our stakeholders or our customers that we really need this infrastructure to be built out to streamline our process of pre-processing the data and not ultimately running our models through it. It's been a huge headache on a few projects just pre-processing the data because a lot of times the infrastructure or the way in which we're collecting this data and storing it is outdated. So that's been a lot of fun.

John Arnold

So what are the ways to remedy that situation with an outdated infrastructure?

Kim Cates

Typically what we've done, and we've had ... I think on a lot of the teams that I've been on, there's been enough funding via R&D. So what we typically do is we'll set up an example pipeline in an environment that we already have accessible such as GovCloud or some kind of cluster set up within an instance within GovCloud more or less and essentially just do a, I guess, alpha initial test case of what this streamlined data pre-processing pipeline look like anywhere from where we're storing the data in some cloud database and then also we're running all of our pre-processing scripts and then loading that in and then we're ultimately training and running our model on it. So that way there's not really necessarily somebody manually storing the data, having some areas occur.

John Arnold

Right. We're going to switch over, change gears a little bit. It's a topic certainly worth mentioning, but as with many industries and sectors in which KBR does business, data science has historically been a predominantly male field. I read the split nowadays is at or around 80% to 20% male to female. How has that disparity shaped your experience in data science?

Kim Cates

Yeah, absolutely. So I always tell my friends I work with all guys. So far I would say that it has been an overall pretty positive experience at KBR. A lot of these guys that I work with, even as some of them, I am their manager, I consider them to be kind of like a brother. That's the relationship that we have. I would say it's overall very laid back. I came from a master's program that was predominantly women, and women, I think, that we're kind of conditioned to be intolerant to failure. I think working in a male dominant field, it's really allowed me to be okay with failing. And that was definitely a shocker for me when I first got into this field.

And then otherwise, I think as a woman in data science, I think that it does require you to be assertive to a degree and really stand up for what you need, as I have been a few times, but also in moderation.

John Arnold

Yeah, sure.

Kim Cates

Because I think that there will be times where there is unconscious bias that comes into play. And people, it's unconscious for a reason, they don't really necessarily know what's happening. There have been a few situations where communication has been a bit odd or maybe things were offered to other of my colleagues that weren't offered to me initially, but I never really assume it has to do with gender, sex, or anything like that. I just address it pretty objectively. Overall, I think it's worked out in my favor. I think that me being assertive and speaking up for myself has actually allowed me to gain more respect among my team and the people I work with.

John Arnold

That's excellent, especially knowing that one of the big things over the past couple of years that KBR has been promoting is that idea of psychological safety where people are encouraged to bring those ideas up and to voice their opinions without any fear of retaliation or anything like that. So that's encouraging to hear that it's been an overall positive experience and that when you speak up, you're listened to.

Kim Cates

Yeah, absolutely.

John Arnold

Well, I’m interested to know what, in your opinion, is the reason behind that and how can the collective “we” remedy the underlying issues? I know that to promote education in STEM fields, KBR does outreach in a lot of different areas. So I’m wondering what we can do to remedy that 

Kim Cates

I think the biggest thing that I've experienced through talking to some of my other female friends that are interested in data sciences or in some other similar fields is this idea of imposter syndrome as if they don't have the credentials or they don't have the background in math or in programming, so they don't feel like they can achieve going to this heavily tech field.

But I think something I've learned in working in this job and also what I see with others too that are more junior, is that it's really as daunting as you make it. Like I said, there's so many resources out there to learn things such as signal processing or some kind of mathematical equation. You can learn it if you really put your mind to it. And I think a lot of people kind of stop there. They're like, "This is too much for me. I'm not familiar with it." So really trying to encourage other people that they can just because for me, I come from a psychology background. Obviously a lot of people have a lot to say about psychology backgrounds, and that's not STEM, what pure STEM is. But I always love to argue that that point of it doesn't necessarily matter, your background, because you can achieve anything. If you really want to you can put your mind to it.

John Arnold

Absolutely. That's an important message for anyone.

Kim Cates

Yeah.

John Arnold

So what is your sales data science pitch to young professionals considering career choices, particularly females?

Kim Cates

Okay. All right. I would say if you love to continue learning and you are confident in your ability to independently learn things on your own and love having the freedom to explore different models and approaches and also creative and analytical at the same time, I would say go for it. It's an insanely demanding field right now. There's a lot that it has to offer with a lot of different applications. And just don't allow yourself to get scared by all the complexities and nuances of the models and the technologies that are out there because there's so many.

John Arnold

It sounds like that sort of answered the next question, which was, what advice would you have for young professionals in data science?

Kim Cates

Yeah, yeah. I mean, yeah, that's basically it. I mean, it's really just a matter of — and this can be applied to anything when you're learning something new — I mean, you really just have to make sure that you remain focused on what skillset you have or what preliminary background and education you might have and what you can reach. Obviously, don't start comparing yourself to other people, because I mean, a lot of people that I work with have backgrounds in computer science or have been programming since they were in high school. So you just have to continuously remind yourself of what your background is and where your goals are in terms of improving for yourself.

I guess that kind of goes into don't allow yourself to get caught up in all the complexities of what's out there, because with the neural network, there's so many things that go into it in terms of how you optimize it to all the different model weights, to all the different architecture. You just focus on one thing at a time, learning one concept at a time, because each individual concept is not that complicated. But if you look at it as a whole, yeah, there's a lot of moving parts.

John Arnold

Outstanding. Well, before I let you go, do you have any parting thoughts for us?

Kim Cates

I would say just as I've been repeatedly saying, just know when to take the lead and initiative on things. And if you believe that you can achieve something, go for it and just don't let other people tell you that you can't.

John Arnold

I'm going to think on that one for the rest of the day. Kim, thank you so much for your time. We appreciate you being on the podcast with us.

Kim Cates

Yeah, thank you very much.

CONCLUSION

John Arnold

I don’t know what other podcasts you listen to, but I imagine you’d be hard pressed to get insights into cutting-edge data science AND valuable life advice.

You can get it all here on In Orbit!

Hopefully you all enjoyed hearing from Kim Cates as much as I did.

And we want to thank Kim again for talking about the amazing work she’s doing with KBR and for sharing a bit about her personal journey.

Also want to thank Emma, our wonderful producer, for the work she does getting these episodes out to your ears.

If you’re interested in learning more about the work we do in data science, cyber, cloud computing and a slew of other critical areas, I encourage you to go spend some time on KBR.com.

If you’re a budding or seasoned data scientist and you’re interested in what opportunities might be available at KBR, you can also check out our careers page.

And if you like what you heard today or if you have an idea for an episode, let us hear about it at inorbit@kbr.com.

That’s all from me!

One last thank you to all you listeners out there for spending some of your time with us today and for keeping us in your orbit.

Take care.