Dr. Evann Smith’s career trajectory has been defined by taking the most interesting, compelling, and challenging opportunities presented. Over the last several years, Evann, along with a team of skilled data scientists, has built Bechtel’s Big Data and Analytics Center of Excellence.
Evann joins Jennifer Whitfield to discuss:
- Opportunities for artificial intelligence and machine learning to make an impact in the EPC industry
- The “art of the possible”
- Advice for the next generation
Listen to the podcast.
Interviewee: Evann Smith (ES)
Interviewer: Jennifer Whitfield (JW)
JW: Thank you for tuning in for another episode of our BechTech Podcast, a podcast where we highlight the incredible technical specialists who work at Bechtel. I'm Jennifer Whitfield and joining me today is Dr. Evann Smith, lead senior data scientist. Evann manages strategy in advanced work packaging within the Big Data and Analytics Center [BDAC] and brings substantial experience in research methodologies and algorithm development to Bechtel. Evann, thank you so much for joining me today.
ES: Thanks, it's a pleasure to be here.
JW: Let's go ahead and dive in. Can you tell us a bit about your career and how you've progressed to where you are today?
ES: […] I started out as an academic. I have a Ph.D. in political science and quantitative methodologies, and I loved my time in academia, but I ultimately came to the conclusion that it wasn't for me. I wanted a faster pace, more collaborative environment, and more directly applied work. But as a part of my dissertation research, I developed and extended some new natural language processing algorithms for Arabic, specifically in colloquial Arabic, and this led to the opportunity to start a company with my advisor working on natural language processing solutions for mission-critical languages, which is what brought me down to D.C. As that company began to hit a more mature stage, I started to pick my head up, look around for new opportunities, and a recruiter from Bechtel actually reached out to me at that time. It was an opportunity that I couldn't turn down, which was to build a data science group from the ground up and to do so for such a significant company. And by that, I mean [a company] whose portfolio of work, and that's from my perspective, [had such a] range of potential questions and challenges on which I could lend my technical expertise[…]. I haven't looked back since.
JW: That's fantastic, Evann. I know you were recently recognized as one of Bechtel's Distinguished Technical Specialists. Can you talk more about your technical specialty?
ES: I'm a data scientist, which I suppose in a very simple sense means that I answer questions from data through mathematical, statistical, and optimization-based models, and a lot of this falls under the heading of artificial intelligence or machine learning. More specifically, in the field of machine learning, I specialize in natural language processing, in predictive modeling, and in complex or dynamic networks and systems.
JW: How did you develop an interest in your field? What type of training did you do to become a data scientist?
ES: Much of my early development and training came through academia, and I often get the question, how do you go from political science to EPC? But I do think that there's a clearer connection than might be immediately obvious. I am trained as a quantitative methodologist, which means that I develop mathematical and model based solutions specifically to help us understand the extremely messy and complex world of data that's created by humans. So, I come from a field with no clean lab experiments or easily controlled data. And when you think about it, that's really not so different from things like megaproject execution and the complex environments in which we operate. Beyond that, I've spent some time at the New England Complex Systems Institute, and I'm pretty consistently engaged with the field through conferences, publications, and all the relationships I have throughout the discipline. This is particularly important [in] machine learning because it's a really rapidly developing field, so it's always a process of ongoing learning.
JW: And what would you say has been the most rewarding aspect of being a technical specialist?
ES: At its core, it seems simple, but it's really the problem solving. I love the creative aspect of understanding new or even old challenges, but then developing novel ways to solve them to provide us with new insights or capabilities. Things that ultimately make us better, smarter, faster in the work that we do. And to that end, I particularly love the point of connection that I can build with a business stakeholder or subject matter expert where [I] truly begin to understand their challenges and questions, and then they, in turn, begin to understand for the first time the “art of the possible,” and then the reward of delivering those solutions and making their life and their job better.
JW: That's incredible. And what would you say are the most challenging parts of your specialty?
ES: I would say two interrelated things. [First,] clearly understanding the question that we're trying to solve. This is also one of the most fun parts, but a lot of times, people come in with very broadly articulated goals, and it's the data scientist’s responsibility to work with the stakeholders [and] the subject matter experts to really distill those challenges into a series of discrete questions that can be meaningfully, usefully, and consistently answered with the data at hand. And then second to that point [is] wrangling the data to build the models that answer that question. Bechtel has over 100 years of data which is absolutely amazing. But it's been collected and recorded in many different contexts and for many different purposes, and we have to make that data not only speak to each other but really understand what it's saying. And, I often say this and I think it's really important, but every act of measurement is also an act of interpretation. That is, when you record a piece of data, you or someone has decided what to measure, how to measure it; [choices have been made] about what's important. And what it actually captures is a function of those decisions. Or, […] put differently, I think it's paramount that we understand the data-generating process to really be able to understand what the data is telling us and how it can be leveraged in data-driven models. And so, these two things are fundamentally interrelated because they're really just about developing [our] understanding of the problem at hand and the data that we have to solve it[.] Good data science isn't magic. It requires a deep engagement to make the right methodological decisions to get to the right answers.
JW: To your point about the amount of data that's out there with all the major advances in data collection and processing, it feels like technology in the data science space is constantly evolving. How will new technology in your field impact the EPC industry and Bechtel in particular?
ES: [ These] are really exciting times for artificial intelligence and machine learning in EPC, especially because the industry is really just at the start of its journey. There are opportunities across […]all aspects of our business for technology and machine learning to have an impact. [One thing] we've worked on and continue to work on in BDAC [is in the area of] predictive scheduling. For example, to better understand our complex dependencies and our ability to meet milestones, automated advanced work packaging [is used] to provide that consistency and efficiency, not only in workface planning, but to drive that process upstream to really coordinate with early planning and engineering. [There are possibilities] in the area of stakeholder and opportunity identification where we can mine the vast amounts of data that are produced outside of Bechtel to really understand what's coming down the pike. And then also, a new area that we're really pushing is in Asset Performance Management, where we can leverage the huge amounts of data generated from sensors, [the Internet of Things], our industrial control systems, for example. [Here,] I think Bechtel has some really interesting and exciting opportunities given our role as an integrated EPC because we design and we build these things[. We] can start generating and collecting data from the ground up and really build that intelligence into the whole life-cycle of an asset[,] taking us not only through testing and commissioning [ but] being able to offer our customers an optimized asset with things like predictive maintenance already built in. But those are just a few of the things that we're working on now. It's really a wide-open world, and I think we're just scratching the surface of what machine learning artificial intelligence and data science can do for EPC.
JW: […] How would you say you specifically have made an influence in the data science field, and what contributions have you made externally to Bechtel?
ES: […] I've had the opportunity to work [on] and be a part of several really amazing research teams, including […] László Barabási Lab at the Center for Complex Networks. Barabási is truly, I would say, one of the fathers of the field of network theory. He introduced the idea of scale-free networks[, which appear all over natural, technological and social systems,] to the world. Also, as a part of the startup that I was with prior to joining Bechtel, I was developing novel algorithms for mission critical languages for various organizations in the intelligence community, and some of that work was deeply fascinating and influential in that space.
JW: What advice would you give to the next generation, especially those who are aspiring to pursue technical specialties within the EPC industry?
ES: Most simply, I think I would say to be curious and to keep asking questions. [This is particularly true in the field of data science and machine learning because it's so rapidly developing and it's really just getting its legs in EPC. I really think one of the greatest skills a technical person can develop is in the area of communication, so I like to advocate for young people and people coming up to focus on being what I call bilingual, which is to say, developing the skills to speak to technical and non-technical folks alike. One of the hardest things to do, and I believe it takes lots of practice, is being able to communicate complex technical ideas and solutions to smart people […] outside your specific domain of expertise. But, that said, I really believe that our greatest advancements come when subject matter experts from a diverse array of specialties can come together and collaborate [and] when we can really learn from each other to drive new and creative solutions forward. This is when […] we make the biggest gains, but this only comes with clear and open communication.
JW: That wraps up this episode of BechTech. Thank you so much, Evann, for your time today. I know our listeners are going to find a lot of value in listening to your insights [about] the data science space.
ES: Thanks so much having me. It was a pleasure to have this conversation, and I hope that it's one that we continue both within Bechtel and with our colleagues outside.
JW: To our listeners, we will be back next time with another exciting technical specialist, so be sure to subscribe to our Insights blog.
Our people, and their expertise, are essential to executing our work. The BechTech Podcast celebrates technical excellence at Bechtel through insightful conversations with technical specialists. Our technical specialists’ contributions both internally and externally help distinguish us as a premiere partner to our customers.