Jul 11th 2021 | E060: How To Thrive In Data Science? How To Get Hired and Flourish?
Caolan Kovach-Orr leads the Data Engineering & New Technology team for ISO Insurance Analytics, Verisk Analytics. Starting next week, he will be starting a new role leading Risk Data Science for Hippo Insurance. He is a Ph.D. data scientist with 12+ years of experience in big data, analytics, machine learning, and new product development. Caolan combines deep subject matter expertise, technical leadership, and a passion for engineering novel solutions to difficult problems.
Before joining Verisk, Dr. Kovach-Orr was a research data scientist who leveraged high performance cluster computing to investigate and understand how variation affects the risk of collapse for ecosystems threatened by environmental change.
In 2018, Caolan led efforts to build and successfully file the Ratemaking product Risk Analyzer Comm Auto symbols. This product uses Tree Based Machine Learning on traditional variables, like engine size and horsepower, for ratemaking purposes. To the best of his knowledge, this was the first regulator approved product of its type in the US.
Caolan’s LinkedIn profile: https://www.linkedin.com/in/caolankovachorr/
The episode was recorded on July 7th, 2021. This material represents the speaker’s personal views and not the views of their current or former employer(s).
Natalia 00:10 Hello, everyone. This is yet another episode of career talks by Welcome Solutions. In these meetings, we talk with professionals with interesting career paths who share their life hacks with us. If you would like to support this channel, please leave a thumbs up and subscribe. It always helps.
Today, I have the great pleasure to introduce Caolan Kovach-Orr. He has a proven track record in team management, including new and existing hires, predictive modeling, creating actionable insights, team management, feature engineering, data manipulation, new product creation, and research and development in academia and industry.
Great to meet you, Caolan. Thank you so much for visiting us today and great to have you. I’m very curious about your whole story told from your own perspective and all the things that we cannot see in your LinkedIn profile.
Caolan Kovach-Orr 01:11 I’m really happy to be here and thank you for having me. I didn’t want to have an office job and I take this from an office right now. I got drawn towards ecology and natural resource management thinking that I was going to be a field researcher and then once I got into the fields, I realized I was pretty good at math and liked the Computer Science side of everything, and I kind of naturally gravitated towards that. I did my undergrad. A few very instrumental professors helped me kind of navigate the grad school pathway and kind of push me into academia which I think at the time was the right pathway for me.
Then I went to McGill and worked on my PhD under Gregor Fuseman. And you know, for probably the first half of that may be a little bit more, I was dead set on becoming a research professor at some large research institution and somewhere around halfway through, I just became a little bit disillusioned with the career paths that were available to researchers. I mean, you wind up in a situation where you’re years away from any type of job security.
You’re grossly under-compensated for your skill level and grossly overworked throughout your PhD, then your postdoc, and then your pre-tenure professorship. I started thinking about leaving academia and going into the industry. There were a lot of different conversations I had both with my supervisor and people outside.
It was kind of a tough transition and a tough decision to make emotionally because it’s been something I’ve been planning on for maybe six or seven years at that point. But, you know, when you think about how old I was back then, that was a huge chunk of my life and almost my entire adult life.
But eventually, I decided for sure to leave academia with maybe the intention of one-day returning, but I think that was just wishful thinking or something like that. I started applying for jobs after I defended my thesis in 2015. I kind of got a lucky break with my current employer Verisk. In that, there was this position that was well suited for me. And somebody had taken the offer for it but then withdrawn with only a couple of months or a couple of weeks to go before the start date. I kind of snuck in with them trying to fill a position and me being the only available candidate.
It was a situation where I was either going to get hired or they were just not going to hire anybody which I was very fortunate for, because I took the job when they offered it and have had a pretty interesting career here for the last six years. I’ve been here for about a year and everyone in my division quit to follow the leader. Our SVP was going to another company. And so it was just me, and the AVP, and one other person left in the division. There were a lot of opportunities there that were a little stressful but I saw that way, you know, 10, or 15, people leave all at once.
I become the person with the most institutional knowledge or whatever. And so I stuck around through that. And then kind of turned into a high-intensity mode for a couple of months where we were trying to build this product using a GLM. If Commercial Auto rate-making purposes that pricing of insurance and it didn’t work. And I was pretty sad.
I went back and I started to use some machine learning algorithms on the same data set even though everybody said that we wouldn’t ever get it approved by regulators. That kind of kicked off this whole thing. Because once we started using machine learning, we had to start using cloud computing and Spark and Hadoop. I built out a predictive model using machine learning and also built the AWS infrastructure that my division still uses today.
And then it took only a couple of months to build the algorithm for a new pricing model. But then another maybe six months of just building out interpretation, so that we could explain it to regulators and get it approved, and then maybe another nine months to get it approved.
By that point, I started building out a team focused on data engineering and data science, and new technologies to kind of push the envelope on things like machine learning, for rating purposes, or infrastructure, like big cluster computing, Spark clusters, and whatnot, and that kind of thing. It’s been a really interesting time. That team started there. I hired two people. And then we opened an office in Poland and I hired four or five more. Now, we’re up to 17. And it’s been a really exciting time. And just a couple of weeks ago,
I accepted an offer to lead a hippos risk data science group, so hippo insurance, and that’ll be starting in a week and a half. It’s interesting. One of the things that I’ve been thinking about a lot lately is that I’ve spent all this time kind of building up this team hiring these people out and training them. People are a little bit nervous about my departure. I try to explain that my departure is probably a good thing for all of them but it’s a hard thing to do. It’s the truth because it’s the same way I wound up in my position is that other people left and then that creates a vacuum and you fill it, or they can fill it in a way that’s unique and special to them.
Nobody’s going to replace me in the same way that I’m not replacing the people before me. I’m bringing my own skills and perspective to the need of the division and other people will have other skills and perspectives and strengths and weaknesses. That’s even good because if it’s just one person kind of directing a team forever, it tends to get a little pigeonholed.
They’re scaled at my level. As you go higher, you want much more stability in senior leaders. For relatively moderate-size teams, it’s good to have some changeover.
Natalia 09:45 This’s a good way of getting promoted or getting to important positions in the company. It happens relatively often in small companies and startups because there’s a high turnover like everything is still in the making and you may become head of the department just because you stayed for two years or more. That’s the only reason. My first question would be since you studied biology, and your research was in biology, from what I understood, how does it help you today, since you are doing mostly hardcore machine learning, like models?
What was the transition? How did you make the transition from biology toward machine learning? Because there is still a stereotype that to do AI and machine learning, you have to have a hardcore technical background. How did you learn about machine learning? And do you think that every biologist can handle it? Or did you have some special training somewhere early in the process? And, what was your path?
Caolan Kovach-Orr 11:15 One of the interesting things is that when people find out I was a biologist, they assume I was working, you know, like a wet lab with cells and DNA or something like that. I was a mathematical biologist. I was working not on the statistical machine learning side of mathematics, but the different deterministic modeling side. Ordinary differential equations, partial differential equations, and massively parallel simulations of bifurcation and configuration are these complex five equations. There are systems of complex equations.
Natalia 12:01 We could have a lot of common topics because I published a few papers in applied mathematics based on differential equations I was modeling, you know because in today’s mathematics, it’s often the case that mathematicians just specialize in one type of equation, and they know how the granting system works, that you specialize in one type of models or equations and you try to force them into as many biological systems as possible to write as many proposals as possible.
You are using the same kind of methodology to research all kinds of things. We could come up with anything from interpersonal like, this is the science the wrong way, of course, what I’m saying right now what I’m confessing here, it’s my confession, I was doing science in the wrong way. We were trying everything with these equations, everything from modeling, and interpersonal interactions, to modeling some networks in the visual cortex.
Wherever you could find a system where you have some local inhibition, then this is what you need for oscillation. You have to have a system where you have excitation one way and inhibition the other way around. And you have some delay in the interactions. And then this’s what generates cycles.
I should check out your publications because I have a gut feeling that I might understand what you were doing. Maybe it would be interesting for me but that was just a digression. I get it now. You have a background in mathematics and you could program before you started working in the industry.
Caolan Kovach-Orr 14:06 I think you’re asking two things which are how did I get to where I am given my background and at McGill and a lot of other Canadian universities. In many universities around the world, the graduate work isn’t almost entirely based on research. I think I took two courses. In my first year of grad school, that was everything. While I had some background in mathematics and computers from undergrad, it was not like having very high formal degrees.
There was a lot of reading and learning and trial and error. There seem to be some cases, some trends in the market for people to write data science blogs to do some type of self-promotion or try to build up their resume. They seem like some sort of leaders. And there are good ones out there. Some people write mathematically sound advice and instructions.
But what we often see is that people tend to extrapolate general trends from these kinds of very specific scenarios. Or maybe I can still remember something about this last night, I was in a bar a couple of years ago, and these guys had just graduated with a master’s in data science from one of the universities in New York, I won’t name it.
It was one of the very good universities and they were talking about how their new employers use window machines and that their new employers were idiots. They’re using Windows for data science. They should be using Mac and these kinds of ideas that you have to do it one way or a day. I don’t know, two months or a year are very detrimental to data science because the big thing about data science is that it’s both Computer science and Science.
It’s truly the science part that’s important to become a good data scientist. And I think that gets often overlooked. People think that I can build an extra boost model. I’m the best data scientist out there. And there’s just a lot more to it than that. There’s a lot of nuance and complexity to what works.
Natalia 16:46 This’s an important subject. We’ll touch it right now. Because it’s not only that professionals publish about data science today to get popular but I think they often feel compelled to do so because you have to become somewhat visible online, especially if you’re freelancing, or if you feel like you’re a beginner in your field, you have to build your name. And I think it becomes more of a standard today that even if you’re employed in a company, you still have to take care of your online image.
You have your personal website, a well-kept LinkedIn profile, some activity online, and some discussion going on. We single-handedly become companies. That trend will progress. The job market is becoming more and more of a soup. We become more and more mobile and become more and more independent, which is kind of counterintuitive because, at the same time, we are becoming a part of all these different tribes online and offline. But it’s like two processes going on time.
At the same time, we want to be acknowledged as personally. We kind of also feel that we have to do it. This’s a part of professional development. I think this’s a part of this phenomenon that it kind of becomes a default. And I think another part of the problem is that, for companies, especially for businesses, it’s very important to get organic traffic.
And for organic traffic, you have to have content. You have to have a CEO. And it’s even there was some research on this and showed that companies that have a blog on the website, sell on average 60% more merchandise than companies that don’t.
This’s because of organic traffic from Google. I’m kind of lucky because I’m a content creator myself, so I can write my own blog. But otherwise, if I didn’t have those writing skills, I would be in trouble. Because then your company website is invisible to Google machines. And every single time you want to attract anyone to your website, you have to pay for Google ads, and this is sometimes a few dollars for one click. This was incredibly expensive.
That’s why today, pretty much every company that is active in consultancy in machine learning and AI, no longer has a choice to be visible to potential clients. They have to invest in content. Why there is such an abundance of content today? It’s forced by the rules of capitalism that you have to do it.
I have to say that last year we had a guest Alikum Lujan here on the channel, he calls a little business, based in Vancouver here in the Netherlands. He’s fantastic. He’s a very good content creator. I can see his content on LinkedIn. I will link it here below.
If you are looking for good content on machine learning, then he’s posting very good and very informational also educational materials that explain to the broader audience how all these algorithms work and what are the do’s and don’ts and how you should clean your data properly. If you are looking for good content in this area, then I can recommend his content. I agree it’s hard to find good quality content among the sea of randomness, especially given how the Google algorithms work.
Google bots are mostly occupied with keywords. They cannot objectively assess the value of your content. They just look up the keywords. They’re not very intelligent. They’re looking at repetitive sentences and phrases on your website. If you have worse content but better SEO, then you will be positioned higher and you will be more visible.
Caolan Kovach-Orr 21:26 It’s funny because we see the same trend in resumes and people who apply to jobs, right? People put these keywords in like Python or Scala. And it gets them more interviews and callbacks. And I think it’s interesting because it’s kind of this tragedy of the commons because every time an employer changes how they do their interviews or how they do their recruitment, the whole market adjusts to kind of match that. It’s the red queen hypothesis where you have to keep moving.
Natalia 22:09 It’s a kind of finite resource. It’s a survival game. It’s like a little moment, we kind of try to survive. I mean, survive that into death conditions. It’s like you fight for life and death anymore but still not enough seats for everyone. There are always more candidates per position than seats. It’s a form of a game. I agree with that. Sometimes, I have to admit that I still apply for jobs mostly because I want to train my own skills in terms of passing interviews, and not because I wanted to help.
It’s just a part of my own professional training on how to advise people with careers. I have interviews. I had one last week. I kind of get skilled in these corporate talks during interviews. But it’s a game and you have to kind of learn some phrases and learn which of the questions are the progressive questions that can get you into the company? Because you can actually sign when you answer these questions. And which of the questions are pure elimination questions and it’s more important.
It’s just about not saying anything stupid because the only purpose of the questions is to eliminate unnecessary candidates. You have to kind of learn this language, like which questions are actually what type? And how to answer the problematic ones.
Caolan Kovach-Orr 24:15 I might phrase it a little bit differently. It’s a game as much as it’s a skill that there are people who are really good at interviewing, who are great employees, and there are people who are really good at interviewing, who are going to not be great employees. And then there are just a ton of people who are terrible at interviewing but would be great employees. Trying to make that transition into at least being able to do a good interview is probably an important skill people need to work on coming out of academia.
Natalia 24:50 As you mentioned before that you also took part in the recruitment process for your employer. I am very curious about what is your personal strategy for hiring people? And what are your principles? And what is your approach?
Caolan Kovach-Orr 25:12 I’ll just give you a high-level overview of the process. What we do is, every time we open up a wreck, we’ll have maybe 200 to 2000 people apply. We’ll screen a lot of them out based on resumes right away, probably 50% get just screened out because they have no coding experience or they hate bosses or something, you know, like silly things that you’ve probably shouldn’t do.
And then we’ll take them and the next step would be a hacker rank assessment. Now, I tried to use that tool as a pass-fail. It’s not about finding out who’s going to be the best employee or the worst employee. It’s about just ensuring that they have the day one skills that we need to be able to do the job. And we’re not going to have to teach them how to do SQL Joins or a little bit of Python. They have all the skills that they said they do on their resume.
I tried to do panel interviews where everybody can veto. That’s not exactly what happens. But we’ll have a bunch of different interviewers having their own specific time and everybody has veto power over a candidate. But then, you know, unless somebody vetoes the candidate, we try to find alignment between everybody.
And it’s a mix of technical and soft skills. Less specific people focus on different things. But I tried to hire people who have the basic skills, and then also have intellectual curiosity, emotional stability, and just they’re willing, you know that they’re smart enough to kind of handle some of the logic problems that will present to them, either in the hacker rank or on like a zoom interview. But I’m not so concerned with someone’s like, if somebody comes from pure software engineering, or comes from the science side, and had a good handle of Python, or SQL, from the science, I don’t care.
I don’t need somebody who’s going to be a senior, true senior-level data scientist to come into the organization. I need somebody who within three to six months is going to start adding value and will continue to add value after that. And that’s kind of for one level. Most of our employees are going to come in at that entry-level. And by entry-level, I mean, people who have masters and PhDs, and maybe a year of work experience, which is a terrible thing for the job market. But that’s considered entry-level. But it’s where we are for higher levels.
And this is something that is often lost. I think a lot of younger data scientists and a lot of younger people coming out of academia see that they can move jobs every 18 months and take a 20% raise. And they do that for a couple of cycles. And by the time they’re on their fourth job. They are making a lot more money than they started with. But they also have never accomplished anything, because they never stayed anywhere long enough to not only learn local systems but also see a project through deployment. They might have a couple of weeks of smaller projects but nothing big in the huge amounts of revenue.
And so what I would suggest to people who are starting their career is to find someplace. If after 18 months, it’s not working out and you don’t think you’re going to add value there, then definitely move on. But if things are going well, even if you could take a big raise to go somewhere else, stay for long enough to be able to add real value. Because when you do that when you add the real value, then you increase your net worth on the open market.
There are a lot of people out there that have made their entire careers off one or two big projects as a PhD or somebody coming out of a master’s. That’s where I think people really want to be. You can handle one or two big projects and become a very senior leader. Don’t leave before you get any of those under your belt.
Natalia 30:15 I was recently going through biographies of the most successful managers of today in the IT industry, including people like Marissa Meyer, and what was his name, Tim Cook, and many others. When you look at the biography, it’s so much money. Somebody else biography is not that she was working in Google between 1999 and 20.
And then whatever it is that we asked, she entered Google at some point but then she took part in designing Google Arts, Google Doodle. There is a list of important projects that today are very successful and that she managed or was involved in as a programmer because she’s also an excellent programmer.
She was doing hands-on work, developing these products. And so she was there when it was deployed and until the end, she was moving to the next one. She has a long list of products and services that are world-famous and in public news today. It’s not a position. Those are projects that she co-authored and co-created.
And this is what should be your list of achievements or the professional projects that you were either co-creating or contributing to and not just the list of positions. I agree. This mentality should change. Because I think there is a misconception. If you want to get really far, then you should have successful projects in your portfolio.
Caolan Kovach-Orr 32:10 I want to kind of hone in on that because there are two things that you mentioned that I think are worth exploring. The first is that if you leave the project before it deploys, someone else is gonna get the credit. And people just don’t realize that. It’s the way it goes. Then the second thing is that you’re right, no one cares. Titles, especially in tech are completely meaningless and mostly made up. Some lead data scientists have hundreds of employees under them. There are directors. Some senior data scientists and managers have teams of 50, or no one at all.
The position title doesn’t matter at all. People look at things like what have you accomplished, the number of people you’ve managed, and the projects. It doesn’t necessarily have to be just products. They can be hiring out teams. They can be training teams, or maybe your project is retention or internal things that never go out to a consumer but you help the company or working for to optimize sales pitches or something. There are a lot of options out there.
Natalia 33:30 No, there’s a lot of debate right now about what the difference is between data science and machine learning. What is the future like how these jobs will look in the future in 5, or 10 years, what will be the real demand? If you are a young professional today who is interested in getting into data science or machine learning, what are the directions in which you might be interested that have the most potential to give you a kind of path? I wouldn’t say predictable career path because you cannot really predict anything today.
But the most options that we know are you can relatively feel safe that in 10 years, you won’t be replaced by a machine, and you might be assuming that you will not be out of the market. How would you approach this if you were just going to get to every class in programming, what would you choose? And how would you start a career today?
Caolan Kovach-Orr 34:48 There are a lot of questions in there. I’m going to kind of take these in order. Machine learning engineer versus data scientists. Again, I think titles don’t really matter. When I do think, that labels can be useful in some circumstances. If we’re talking about the data scientist who builds machine learning algorithms on and has a data engineer feed them the data and all they do is build the models using open source tools. I do think that that job is in danger of becoming obsolete or at least far less compensated than it is currently, you know, kind of like a gold rush in California in the 1800s.
The people who made the real money were the ones selling the tools. The companies that are selling the tools are the ones who are going to make real money, you know, data robots, who will do all that stuff. We’ve been on a wave, where 10 years ago if you were going to build a neural net, you had to mostly write the code for the neural net from scratch. You were building your own loss functions. You’re building everything.
Five years ago, you get the open-source packages but all the tools and metrics where you need to build yourself, and now those packages are so advanced that it’s very easy to take people with very little experience and teach them to run boosters or whatever. If I was going into data science now, how I live now is I try to focus on the hard problems that no one else wants to solve but that need to be solved. And those kinds of things aren’t gonna go away. In data science or whatever this field is, the science part is often overlooked.
That’s the part that isn’t going to go away, like knowing how to design your experiments, knowing how to design your systems. If it’s prescriptive, you’re going to pass it off to an analyst-level position. They can run that model or do whatever. If it’s something that requires real problem solving, and it’s hard to do, I think that careers are always safe and also well compensated because there just aren’t a lot of people out there who want to take it to that next level.
I would think about things like getting if I was in school, I definitely take courses on big data. Because those principles become even more and more important even when we’ve got fully managed spark clusters and EMR and all that stuff like just understanding the concepts behind how all that big data services work. I think it’d be really useful. I take courses on experimental design, and probably some courses on ethics because I think that’s going to be a big thing over the next 10 years.
Visualization can always be helpful, especially for younger people. I wouldn’t necessarily go into a career and visualization. But being able to communicate your results is important. Taking some real science courses, not wanting to 100 level, bio, chemistry or physics is important. They’re the ones where you’re learning at the graduate level. It’s, you know, four or 500-level course, where you’re reading scientific papers understanding why they did what they did, and then breaking down what they could have done better.
And on those literature reviews, understand why they didn’t do the thing that would have made their study better. Oftentimes, that gets overlooked because people are like, Oh, they should have done it in XYZ way. That would have cost millions of dollars and nobody was ever going to do that. If you can get a sense of what corners you can cut and what corners you can’t, I think that can always help.
Natalia 39:07 There are lots of calamities in the history of engineering and for instance, spacecraft that you can read about that come from a lack of fundamental understanding of the laws of physics where billions were spent on the mission. There was this famous story where this platform in the sea kind of got crashed because someone didn’t try to triangulate the shape of the player like the bottom of the platform properly and didn’t understand how to do it.
And then it all crashed. Billions were lost. This’s for data scientists but this’s like some people argue that this is exactly the reason the difference between data science and machine learning engineering that engineering requires knowing the data. And that’s why it’s actually in the long run, better than just pure data science.
Caolan Kovach-Orr 40:15 It’s funny. If you look at the history, so the highest paying best job in 1985 was analyst, like, that was a person making 300,000$ a year and a 40-hour workweek. There were a couple of 1000 in the whole country in the US. Then people who weren’t operating at that same level started becoming analysts. Then the analysts change their title and their business intelligence and then the people who were not operating at that level, but had gotten the analyst’s title go after that.
We go through this whole thing and it becomes data scientists or data analysts and data scientists and machine learning engineers, and you’re using a label that will only last for a couple of years until the term changes again in our world of people who use machine learning and statistics.
Regardless of the title, there are different types of data science and machine learning engineers out there. In industry, some people do cutting-edge research that builds the tools. They’re coming up with new types of neural nets and they’re writing the source code for different algorithms.
That’s type one. Type two people are using those algorithms and tools to solve new problems. Type three people are the ones who were using those tools to solve old problems and do refreshes and things like that. Maybe they use machine learning for marketing at one company and they’re gonna bring it to a new company. I think the fun is in being a person who’s in category one or two. As long as you’re in that area, you’re going to be okay and that’ll be fine. Because you’ll have a lot of longevity.
They’re the people who are just using the tools to solve the same problems that have already been solved. Those are going to always be the following one step behind, maybe two-step depending on time. I would say, not everybody’s going to become the type one making new algorithms person. I think you need a lot of technical formal training. But there are a lot of opportunities to be the type two-person to take, you know, machine learning and solve a new problem or to find a much better way to solve a currently solved problem.
Natalia 43:05 Many researchers who are watching this channel maybe think that this’s parallel to research because, in research, we also have these fundamental researchers who are working on new methods or new ideas, new concepts, and new mathematics. I was working in neuroscience or new methods to analyze data, versus the experimentalist, the applied researchers who are taking those methods, using them to try to answer new questions.
I think there’s just a parallel thing. If you have three to four experimental applied research, maybe type two is for you. If you’re more of a fundamental researcher, maybe you’re the material for type one.
Caolan Kovach-Orr 43:59 I don’t know how to bring it down. I kind of disagree with you there. Because there are a lot of people who do basic research on who would not be good at type one. There’s like thinking of biologists who might be in the wet lab, who are doing cancer research that it’s gonna be a really hard transition to go from, like cutting edge cancer research to being the person designing new algorithms.
Natalia 44:28 I was just thinking with the assumption in my mind that you can program if you know. I hope you get it.
Caolan Kovach-Orr 44:35 I got you. Now, that makes more sense. I was thinking broader. There’s a strong parallel type of thing.
Natalia 44:43 I agree. You have to provide the value. You have to be able to offer something new. It must be either a new business model, a new application, or it has to be a new method or one or the other or both but it has to be a specific point at which you can provide something new, and build a novelty. That makes it special, I guess. I mean, this is obviously what I’m seeing right now. But this is also tricky because when you look into what the typical job interviews look like, they have phases, right.
In phase zero, when you’re admitted to the whole process, and you have to do with an external recruiter, or you have to do with like this recruiter within the company, but like, let’s say that it’s a corporation, so multiple levels. And this is like this recruiter of the first contact, who is not directly related to your team, it’s just, you know, the first level of selection, then when you come to the interview, you have to kind of show that you’re just as closely fitting the pattern that they want as possible.
You’re nothing special, nothing extra. You just admit to the requirements, as tightly as possible. But then, once you proceed to the next round, so you’re not eliminated because you don’t stand out too much. You fit into the frame that they want, then once you get to the next level of recruitment where you meet with the hiring manager who might work with you later and might even be your boss, you have to show that you can give as much extra value as possible.
It’s often the case that people want to stand out. They want to show that they can produce something extra and that they are so much more than what is expected. But they show it like in the process too early and eliminated, it’s quite a common problem. Because the recruiter of the first contact only has a list of requirements and they only check the marks. This’s tricky. You have to provide value. You have to prove to the employer that you can give extra quality. But you have to know when to use this card.
Caolan Kovach-Orr 47:25 I think it also really depends on what level of job you’re applying for. I talked about this a lot and one of the areas that’s probably hardest for a lot of academics in the transition to the industry is that you are again going from being a big fish in a small pond. Even when you’re a graduate student, you might not feel like a big fish because there are all these tenure chair professors around who are brilliant.
When you’re trying to get that entry-level job, I don’t know if it makes sense to try to shine and show in the interview, like how great you’re going to be because a lot of what the recruiters are looking for, beyond that checklist of like, can you write code in Python, and you know that you are going to be easy to work with. And so if you come off as egotistical or self-important, I think it’d be really difficult to get through to that next round, where you can show how great you are.
And even then, you take that first job out of grad school and try to learn as much as possible and adapt to the things that you didn’t have to adapt to in academia, you know, especially around the soft side of things and what people care about. And then you know, after 18 months, or two or three years, whatever it is, try to go for that big jump because nobody’s going to hire a recently minted PhD for a principal data scientist position. If you do get that job unless you had industry experience going into your PhD, likely, that you’re not working as a principal data scientist. It’s just an inflated title like VP.
Natalia 49:27 You’re never going to be the only data scientist.
Caolan Kovach-Orr 49:29 It can be a great position and a startup, but also I would say that there’s an opportunity to learn a lot, and obviously, my advice is kind of geared towards people who have a similar pathway to me coming from applied mathematics, PhDs and not people who come out of Stanford, you know, physics or machine learning or computer science or something like that.
Natalia 50:04 It also makes me smile. In small companies, everyone is the head of the department. In academia, it’s just you are a 30-year-old professional with a few years of experience and you’re still like PhD candidate. It doesn’t sound so dramatic, you know. And then, in these companies, sometimes five people have all position titles that are like 10 10 words long and everyone is like the Principal Director of the Department. It’s just funny.
Caolan Kovach-Orr 50:46 You see the same thing. People go into banks and finance like everybody’s a VP but nobody cares, right? Like, you don’t fight if somebody applies to one of the positions with a VP title, it doesn’t, am I gonna reconsider it and be like, oh, I should make them a senior manager or something like that. It’s whether or not you have the skills for the job.
Natalia 51:12 Right. I couldn’t just feel serious about it if I had a position like this but I get the point. I mean, when you like it, it maybe gives people more motivation, and they look better on their LinkedIn profile. If you can give them that value for free as an employer, why not give it like, I would also do the same. If I had employees, I would just give them the best position titles they can have. As long as it’s not the owner, it’s fine.
Caolan Kovach-Orr 51:49 If I think from a management perspective, I try not to do that. Because I want to match up people’s jobs with their titles in a way that’s realistic, right? Because it’s not fair to ask somebody who’s operating at a VP level to take a director title. In the same way, it’s not fair to ask somebody to take a director title when they’re operating as a junior data scientist because it’s just setting them up for failure.
I mean, one thing we see all the time, is people get these titles. And then they go on Glassdoor. And they’re like, Oh, well, I’m a senior data scientist, I should be making 40% more compared to all the other senior data scientists in New York, except those senior people with that title at other companies tend to be much more senior. And so you’re just setting up the employees for poor morale because they’re comparing themselves to people with 20 years of experience.
Natalia 52:52 If I can say something about it, I have to say that this is one thing I like about business development. If you have a company, nobody cares how many years you spent in the business, like, when the client comes, they want to have a job done. I started my company two years ago.
And I do like workshops, training, public talks, all kinds of like services for resources at the moment, but I’m also planning to launch services for companies. And I’m serious there. There are other people who also I know, they’re too good at what they do. But they are in the same space for 10 years and longer.
And if I feel I have content that is the same good. I charge the same. I will not start half just because I’m here for two years. It matters how much I can do and what is the value of the content. It doesn’t matter how many years I was here up to this point. And the clients respect that they want the value and they pay for that.
They don’t pay for how many years you were doing the job. And that’s what I love about having your own company, it actually matters. You have to sweat, how much you sweat, how much you can give that matters. It doesn’t matter how old you are or how many years you were around. And this is also one thing that kind of keeps me away from entering the corporate world as an employee. I never even had that attraction to do it.
Because I know that you have to prove yourself also being in the field and being in the field is also valued. I would have to build up to something for like 10 or 20 years to be able to say Hey, I can do it best in my field because I’ve built this portfolio of projects and positions and I established some authority.
In business development, time passes by differently. It’s like different dynamics. It’s all about how much your client values your work. That’s, you know how much you can give. I mean, nothing is wrong with either of these. I’m just saying that they’re different. And I like the entrepreneurial way more but it’s just my feeling. But of course, there are pros and cons.
Caolan Kovach-Orr 55:31 It can be challenging to prove yourself, especially coming out of academia, where it’s depending on your employer’s experience with academia, they may be thinking of people who have masters and PhDs, as people who did an extended undergrad and didn’t want to enter the job market. From my own experience, it’s not at all what graduate school was.
You can come out of this. When you enter the job market, it can be a little jarring because people start treating you like, Oh, you’ve just graduated from college, even though you’re maybe in your late 20s or early 30s. You have been doing research for 5, 6, 7, 8, 9, or 10 years. And so that can be challenging. And it is an uphill battle. The only thing I’ll say is that, if you’re at a good company, it takes maybe 18 months for people to kind of get the message and understand who you are and for you to build that reputation.
But after that, it starts to fall into place pretty well. There can’t be that opportunity. It doesn’t take 20 years to build a reputation for people to respect you. It might take two years and even then some people could do it faster than that.
Natalia 57:06 I think we’ll have to slowly wrap up. I’d like to talk to you more about maybe some other time, you know, there will be other locations to talk on this channel. But let me still ask you one question that I often ask, is there anything in your career up to this point that if you had a chance, you would do differently? If you could turn back time five years, what would you do differently with your career?
Caolan Kovach-Orr 57:40 I don’t know if I would do much differently. I wish I had done it earlier. Verisk has helped with this quite a bit. I’ve got some great people I worked with at Verisk. They also do a lot of support, like employee development through these kinds of official horses that they run through Harvard Business School.
I think some of the things I learned in those courses I wish I had known before. When I started, those should have been like day-long classes. And it took a little bit of time to learn. That might have been something that I would change, but you know, types of things that you’ve learned there are that when you’re in academia, the most important thing is that publications and grants.
That’s a success. And when you’re in a company, what’s most important is not your own success but the success of the company and helping other people. Everybody says that but I think a lot of leaders also look for those qualities and value those qualities in other people. And I don’t think when I came out of academia that I fully understood how important it was to help other people and how their success is much more important than my own success. That’s the big thing.
The whole management structure isn’t designed around some system of like, gates, where each person up is just in charge. It’s designed around, who is going to motivate people, who are going to help them do their work, and get things accomplished. And those are very different skills from what we might value in academia. I wish I had known a lot of those types of things coming in or at least quickly after I started because I think it would have made the first couple of years a little bit easier.
Natalia 1:00:00 And I think we still have that reflex, right? We have that natural reaction after so many years working in academia that we have to first and foremost, build our own CV. It’s hard to change the mindset they want. It takes time for sure. It’s very important what you just said. Okay, great. Thank you so much, Caolan, for all this great information and all your insights. It was great talking to you. And for all of you guys who would like to get more of this type of content, please subscribe to this channel. Leave us some thumbs up.
And if you have any questions for Caolan, then please contact him through LinkedIn or post your questions below. We will answer each one of them. You’re welcome. We’re waiting for your questions. And thank you so much for accepting our invitation and for being with us today.
Caolan Kovach-Orr 1:01:06 Thank you for having me. This’s great. And please let me know about any future panel discussions.
Natalia 1:01:12 Sure. I think panel discussions are not super nice. I think I cast some of them. I think it’s a good idea. Thank you so much, guys, for watching.