From Ecology to Data Science

This interview first appeared on The Accidental Engineer.

In this conversation, Zach Deane-Mayer, Director of Data Science at DataRobot, shares his journey from studying ecology to becoming a data science leader. He discusses the value of learning programming for automation, emphasizes engineering skills as crucial for data scientists, and recommends practical ways to break into the field - including applying data science to your current job and learning Python. Zach also explains how DataRobot automates machine learning while highlighting that the most valuable data science skills remain human-centered: communication, problem-solving, and understanding business needs.

Listen to the full interview:

Read the full transcript below:

The Accidental Engineer

Patrick Wong: Hello and welcome, I’m Pat of The Accidental Engineer. And today on the show, we have Zach Deane-Mayer, Data Science Director at DataRobot in Boston, who is also the creator of a few data science courses at DataCamp. So Zach, thanks so much for being here. Could you start by telling us a little bit about yourself?

Zach: Yeah, I’m Zach Deane-Mayer. I’m a Director of Data Science here at DataRobot, where I’ve been working for about three and a half years, building automation-focused data science. Prior to this, I worked at another startup here in Boston, also doing data science and machine learning. Before that, I worked at a management consulting firm doing data science. And then before that, I worked in energy consulting. So I kind of got into data science by a somewhat odd path. I actually studied ecology and mathematics in school. And then kind of my first job out of school, I got into an econometrics job at an energy consulting firm doing like linear regression modeling for electric load forecasting. And so it wasn’t really called data science back then. Like it was just, a job out of school.

Patrick Wong: Yeah, that’s really interesting. At that time, data science was just gaining popularity as a term. I believe it was coined, sometime around 2009. But yeah, that’s a really interesting path. Can I ask what led you to choosing to study ecology and math?

Zach: Yeah, So let’s see. I took biology in high school and that was my favorite science class. And I was always interested in science. And so I said, OK, let’s major in ecology in college. Like I took some ecology classes. I enjoyed them. And that’s why I decided to study. So I wasn’t like really thinking about a job or a career after college at the time. I was just taking classes I enjoyed and kind of ended up majoring in a topic where I’d taken a lot of classes. And actually kind of, it was sort of funny. Like after college ended, all of my friends had gotten jobs before they graduated. And that hadn’t even occurred to me as a thing to try to do until, I had my diploma in my hand and thought, what’s next?

Patrick Wong: So what did you think about your career when you in your first role when you mentioned you were running a bunch of linear regressions?

Zach: So at the time, so I’d taken a few stats classes. I actually minored in math. And basically I moved out to San Francisco with my girlfriend at the time, my wife now. She had a job lined up with Teach for America in Oakland. She kind of had, her next two years planned out. And I had no idea what I was doing. And so I did a couple of like odd internships and random things. And I kind of just spent like three or four months looking for a full-time position in San Francisco. And I eventually found one at a consulting firm where, as lots of consulting firms do, they hire people from, somewhat quantitative backgrounds. So I’d majored in math or minored in math. So that kind of counted to do data analytics projects, basically. And so this consulting firm I happened to end up at was an energy consulting firm.

Patrick Wong: Cool. So what’s an example of one of the projects that you worked on at that time?

Zach: So we would do electric load forecasting for electric utilities. So we would get like, five to 10 years worth of hourly data. And we would produce a forecast that would say, all right, for every hour of the next year, here’s how much, electric load we expect this utility will need to meet. And we do that for different scenarios. Like, what if it’s a really hot year versus what if it’s a really cold year?

Patrick Wong: Got it. Well, that’s definitely a perfect application of data science. How did that lead you into your role at DataRobot today?

Zach: So after we lived in the Bay Area for two years, my wife and I moved back to Boston where I continued to work for energy consulting firms doing very similar kinds of work. However, a lot of those consulting firms I looked at, I worked at, were very focused on like traditional econometrics and statistics. And so around the time I moved back to Boston, I also started competing in machine learning competitions on Kaggle. And that’s where I would really say I first kind of got my first taste of like real data science and first got introduced to machine learning. And I loved it. It was amazing. It was fascinating. It was like a completely different way of looking at how to solve problems with data. And so I started doing a lot of Kaggle competitions in my free time outside of work and learning these, languages like R and Python and the machine learning packages in them. And eventually went on to get a job using that. But so I definitely I started out like outside of work, learning it on my own just because I was fascinated with it. Eventually, I started pulling some of those skills into my job at the consulting firm. So eventually, at the management consulting firm I worked at, I actually was doing data science because I was taking the skills I’d learned outside of work and trying to apply them to the problems I was facing at work.

Patrick Wong: That’s really cool. And that must have had a huge impact on the work that you were doing.

Zach: Yes. So early in my career, I took the time to learn a programming language. And I think that’s that’s more expected of like people in an analyst level now than it used to be. But, at least 10 years ago, that was a pretty big competitive advantage for me because, even then I was very interested in automation. Like I think automation is a common thread that runs through my entire career path. But even then, I could figure out ways to take a problem that says solving with, and I think everyone’s familiar with the giant Excel sheet with with like, hundreds of thousands of formulas. And it’s like, following one thread through seven different sheets can be can be an experience.

Patrick Wong: Absolutely.

Zach: And I had a lot of situations where I could take an Excel sheet like that, sort of think about the big problem it was trying to solve, like the calculation it was trying to accomplish, and then rewrite it as say, 50 lines of R code that would do the same thing. But in a more modular and repeatable manner, so you could take, the R code and tweak one part of it and look at the output and then go back to the old version of it, fairly easy. And, and, and, and like, it gives you, it gives you a better way to automate a lot of those data science workflows. And so at the time I started, like everyone, there’s a lot of, of, of analytical work that was being done in Excel. I think that’s less true now because, more people are realizing that, that programming is a better way to solve these sorts of problems.

Patrick Wong: Yeah, definitely. So you were essentially using programming in the beginning to automate your job. You made your job more efficient and then that kind of opened up more time for you to do more interesting things.

Zach: Yes, absolutely. And so that’s the thing is, I’d get this Excel sheet and my job would be, all right, you got to wait a week to fix the calculation. And when you replace that with an R script, it’s not like you’ve got a week to fix the calculation. It takes you five seconds to change one line of code. And now the calculation is updated. And so that frees up a week of your time to, instead of fighting with Excel, focus on the real problems of your job. So for example, we’ve updated the calculation, but the results still don’t look right. Let’s think about the problem we’re trying to solve or the data we’re putting into it, or what does the client really want here? They don’t just want an updated calculation. They want a new way of thinking about this. And so it’s like you take a lot of that drudgery out where you’re fighting with Excel and you can focus more on is my client happy? Or if you’re an analyst in a new job, is my boss happy? What’s the problem we’re really trying to solve here? What can I add as a human instead of an Excel monkey?

Patrick Wong: Definitely. I think that for myself in my career, that’s how I found so much of my interest to develop in programming, just seeing how valuable it could be in terms of just automating all this work that I didn’t want to do in Excel.

Zach: Yeah, exactly. Exactly. Like it really sucks. It’s boring.

Patrick Wong: Yeah, So it just made my life so much easier. And then I had time to do the really interesting stuff, some of the more advanced modeling that helped us solve problems and a much more developed framework. So I think that was really valuable and a really great entryway into the field.

Zach: Yeah, I would, I would totally agree with that. And honestly, I think it’s a good way to think about the field right now. Like, like, I, automation in data science, is best applied to the sorts of problems that humans aren’t even very good at anyway to begin with. and it’s a good attitude to have, like find the boring parts of, what you’re doing at your job or what your company is doing and figure out a way to automate them so humans don’t have to do them. And that like, that frees up those humans to work on, much more interesting and also, much more challenging things.

Patrick Wong: Definitely. And that’s, that’s sort of what DataRobot does, right? So for those who aren’t so familiar with DataRobot, can you tell us a little bit about what the company does?

Zach: Yeah. So we’re a data science automation platform. We build a product called DataRobot. It’s a web app and it basically does automated model fitting, model search, model validation. It, does grid search and tuning for all your models. It does ensembling. But it kind of, it takes like all of the best practices for predictive modeling and machine learning. And it automates them to the point where you can put in a data set, select your target and get out the other end. Very good models.

Patrick Wong: That’s really cool.

Zach: It’s, it’s, it’s an incredible product and it automates, all of the parts of machine learning that I wasn’t very good at as a data scientist, like, remembering which parameters of XGBoost to set to what settings to get the best results. And what DataRobot kind of does is like, it remembers that, but for all of the algorithms. And so, you might be an XGBoost specialist, but then when you hit a problem where a linear regression model is really the best approach, you might not quite remember, the best practices for representing data for a linear model versus an XGBoost model versus something else. And what DataRobot does is you can kind of take all of that implicit knowledge that, different specialists in one kind of model have and put them all in one place, basically.

Patrick Wong: That sounds like it’s really useful.

Zach: It’s incredibly useful. And what it does is it lets you as, not even a data scientist, as a person with data who has a problem that they want to solve, it lets you focus on what data you have and what the problem is you want to solve. And so rather than spending like many months building one model, DataRobot lets you build many models a day. And so you can look at a data set and say, cool, DataRobot can build a really good model for this problem. And then you can think about now, how do I operationalize that model? How do I take this model and make decisions from it? Like what kind of business process can I build around this model? And that’s a much bigger and more challenging problem than how do I build a model? And it’s honestly the problem that as a human, you need to be thinking about, how does this model impact my business? And no computer can answer that for you. You’re the person who’s the expert at what you’re doing in your business. And you should be thinking about models as a tool for changing your business, not models as an end purpose in and of themselves.

Patrick Wong: Right. So it sounds like the DataRobot product does a lot in terms of automating tasks that a data scientist may not want to do or certain things that a data scientist might not be as great at doing. So what would you say to somebody who is concerned that maybe automation might kind of take over certain data science jobs?

Zach: Um, yeah. So what I would say to that person is, especially if you are thinking about what skills you need to learn to prepare yourself for the future. I say to all of those people and everyone I talked to who’s interested in data science that you really need to be learning engineering skills. I see data science is very much moving towards an engineering discipline. And people who have programming skills as we’ve already discussed, people who have an engineering mindset, people who like to build things. Those are the kind of people who are going to be really successful to my view in the future of data science. And so I think focusing data science very narrowly can be dangerous because the world is eventually going to shift out from under you. But if you think about data science broadly, data science as a way of building valuable things, there’s always going to be demand for that. I really don’t see that ever going away.

Patrick Wong: Got it. So what would you say to a data scientist who is looking to learn some engineering skills? What sort of specific skills should that person be looking to improve on or learn?

Zach: So I tell everyone to learn Python, there’s this kind of R versus Python debate out there. I learned R as my first language. I really loved it. I still love it. I still do a lot of work in R. Ultimately, I think looking to the future, Python is going to be a really useful language to know. And one reason for that is especially early in your career, it limits you a little bit less. Like there’s a really wide range of jobs out there, including non-data science jobs where Python is an incredibly valuable skill to have. In five years from now, my own career was in many ways, a discovery process about what kind of jobs were out there and what kinds of things I was good at. And when I started working, I really had no idea where I was going to end up. I had to figure that out. And so limiting your options in the future is not a great idea. You should be thinking about what are tools that give me broadly applicable skills to many different kinds of jobs. And Python is more of a tool like that. I always recommend people to learn Python. And then I think on top of that, there’s a lot of the softer business skills that people sometimes neglect to think about, but just things like public speaking, presenting, being able to communicate effectively. Whether you’re a data scientist or a software engineer or a project manager or any other kinds of positions that you could aspire to have one day, being able to communicate effectively is incredibly important.

Patrick Wong: And that’s definitely something that automation can’t take over. It’s, uh, those soft skills that, uh, a data scientist really offers value in terms of being able to solve problems and communicate them to different audiences.

Zach: Right, right. So like, anybody, I mean, not anybody, there are many, many people out there who can build good predictive models. But, I used to work at a consulting firm and a big part of what we did was predictive modeling. But that was never the hard part of the job. The hard part of a job was figuring out how to use the model effectively, and then convincing all of the different people in the IT organization and in the business organization and all the other organizations that the solution you built was the correct one. And going and finding those naysayers who are like, no, this machine learning thing isn’t going to work. And figuring out what they didn’t like about it. That’s another way of thinking about this. There’s a lot of people out there who see data science itself as a threat to their jobs, so data science automation. It’s the wrong way of looking at it. You’ve got to think about what are the people out there who aren’t data scientists, and how does what you’re doing impact them? And how can you make sure that the model you’re building, or the product you’re building around the model is going to serve their needs as a non data scientist?

Patrick Wong: I think that’s really great advice for aspiring data scientists. I think that that’s something that you can do in your current job.

Zach: Yeah, exactly. Exactly. So like, let’s say that there is some project that involves a boatload of data entry at your current job and nobody ever wants to be the data entry person. And if you volunteer and you just buckle down for a couple hours every single day and you crank through some data entry problem and build suddenly a valuable data set that didn’t exist without your hard work. You’ll have built up a little bit of credibility when you say, hey, and I want to do an analysis of this data set or here’s my analysis of this data set or here’s a cool model I built while I was cranking through all this data entry. And people will appreciate you doing something that no one else really wanted to do. So that can be a way to at your current job start to pull in, if you look for opportunities, some of these data science skills that you want to be using more at work.

Patrick Wong: Yeah, absolutely. Data science courses online are really amazing because they provide you with all the data on hand and, you get to get your hands dirty and start modeling right away. But the thing about that is that that’s usually not how the job works. There’s a lot of data cleaning that’s involved before you get to do the fun stuff.

Zach: And it’s not just data cleaning, it’s getting the data. It’s going down and finding Jen, the mythical person who you’ve never met, who manages some inventory spreadsheet that’s got all the necessary information. And convincing her that you’re a trustworthy person to email her life’s work in the form of an Excel file to. And there’s all that kind of human relationship stuff that goes into just getting the data that you need. And it’s like, yeah, the modeling isn’t the hard part of the job.

Patrick Wong: So you mentioned your DataCamp course a bit back. Could you talk a little bit about what DataCamp is for those who aren’t aware of it?

Zach: So DataCamp is an online education site and they basically teach mostly data science focus, but also kind of general programming classes in both R and Python. And it’s a it’s a very cool model. So they have you write code. Most of their classes are very focused on, giving you some instruction by a video and then and then very quickly moving to a set of coding exercises where you write some Python code or R code. And then you get immediate constructive feedback on what’s working and what isn’t in the code that you’re writing. And really, like the best way to learn how to code is by writing code. And DataCamp very much focuses on the act of writing code with, supervision on top of you that that looking at the code you wrote and helping explain to you why it isn’t working or it is working. So it’s a really good place to learn like the basics of R and Python to get over that initial hurdle of like like how do I code. And so I’ve I’ve actually got two DataCamp classes. So I’ve got an R class that I’ve had on there for a couple of years. And then my Python class just launched at the start of this year. And the Python classes is really it’s kind of a labor of love. Like I talked about how much work it was to build earlier. But it’s on the Keras package, which is a deep learning package I really, really love. And it’s just cool to me because you can take a four hour class and it’s just so easy to do some of these really innovative, completely novel, deep learning things. And like you hear about it all the time, you read about it all the time, you can go through not too many DataCamp courses and like actually have an idea of how to do it yourself. And like that’s really cool to me.

Patrick Wong: That’s really awesome to hear. I think that I would definitely be interested in taking that class myself. I haven’t used Keras at all, but I know that it’s a it’s a really awesome tool. I’ve only heard good things about it. So why would why would you encourage someone to take that class?

Zach: So I would encourage them to take the class because one, I get paid if you take it. And that’s great for me. But two, I really love Keras. It’s a very elegant way of defining neural networks. I did not understand deep learning or neural networks until I started using Keras. It’s a super intuitive way of building those kinds of models. And then that really helps you understand how they work. And so even if you’re not going to be doing deep learning as part of your job, I see I still think it’s valuable to have an understanding of how these models work and how they think about the world. And like there’s nothing that’s magic, within deep learning. I think there’s an XKCD joke about how deep learning is just a pile of linear algebra. And it’s really true. Like Keras kind of strips away all of the mystique and craziness you hear around deep learning and just shows you how simple it really is at its core. And you can solve some really interesting problems with that. But, what I love about Keras specifically is it’s it’s a really flexible way of defining models that are quite specific to the exact problem you want to solve. And so you can do some really creative things with it that are harder to do with other types of machine learning models. So I love it. The course is called Advanced Deep Learning in Keras with Python. Check it out if any of that sounds interesting to you.

Patrick Wong: Yeah, that’s awesome. I mean, I’m sold. So that’s great to hear.

Zach: It’s also a four-hour class, so it’s not exactly a huge time commitment. And then you can email me with what you think of the class after you take it, and I will respond to emails about the class. I’ve gotten a few already.

Patrick Wong: Cool. So how much deep learning do you do in your job on a day-to-day basis?

Zach: Not a ton. Most of my job is focused around engineering. And the kind of funny thing is, is a lot of the, engineering problems I’m trying to solve aren’t even directly related to data science. But, as we talked about earlier, they’re necessary problems to solve because I have a data science product that I want to deliver to a market. We have deep learning models in DataRobot. We’ve actually got a team of deep learning scientists who work on them. I would say they do a lot of deep learning as part of their day-to-day work. They’re much better at it than I am. And I mostly just, look at the problems they’re solving. look at the code that they’re writing. I’m not writing a lot of it myself, but there is that work being done at DataRobot. I pay attention to it because I’m interested in it. I wouldn’t consider myself a great deep learning scientist, but, I’m good enough at it that I can teach a course on Keras. I’m maybe not good enough on it to like, build another ResNet or something like that.

Patrick Wong: And what do you see, what do you see deep learning really providing the most value for practitioners out there?

Zach: Yeah. So, see, you talk about data science, like sort of like signal to noise ratios. Deep learning, deep learning really excels at problems where you have lots of signal, like where you have like an overwhelming amount of signal where you’re like drowning in signal. So, so where you’ve got just like a lot of information that, and you’re trying to make sense of it. So for example, images have a lot of data in them. You look at an image of a dog and there’s not a ton of noise in that image. Like maybe 60% of the pixels in that image are of a dog. The problem is there are like a lot of different ways that a thing can look like a dog and you can collect millions of examples of non-identical dogs. And like it can be overwhelming to look at all that and say, whoa, how are all of these things similar? So deep learning is really good at taking a problem like that where there is like more signal than you know how to deal with and finding a way to structure it and make sense of it. Deep learning is less powerful in areas where there’s a lot of noise. So for example, there’s anyone, especially people who are just learning machine learning always go out and try to predict the stock market. Unsurprisingly, no machine learning methods are that good at predicting stock movements and deep learning models are especially bad at that as fun as they sound to try to do.

Patrick Wong: Got it. So for any aspiring data scientists out there, you would recommend that they kind of focus on some of the more fundamental machine learning algorithms before really diving into deep learning.

Zach: Yeah, that’s generally the advice I give. I think deep learning is a good tool to have in your toolbox. I think there’s a lot of applications where it makes sense. In general, I say to people like, like try to become a good engineer, try to learn Python, try to get good at Python. I would say if you’re really passionate about things like image recognition or if you’re really passionate about, there’s some exciting things going on with language data right now in deep learning, it’s never a terrible idea to follow your passion. If you’re not especially passionate about, say, images or text, you want to be more of a generalist, it’s probably safer to learn the sort of more traditional machine learning frameworks like Scikit-learn. But I mean, honestly, it’s a good idea to start with the basics. I don’t think there’s much harm in that. And, I don’t know. Deep learning is really fun. There’s a lot of passionate people out there in it. Regardless of what you’re doing, whether you want to be a generalist data scientist or a deep learning practitioner, if you’re going to be successful five years from now, you also need to be a good engineer. And so like that’s what I tell people to kind of focus on first, like learn your Python software engineering skills and then apply it to like the realm of data science that you’re most passionate about.

Patrick Wong: So I’m assuming that you also help DataRobot with hiring for data scientists. And so I’m hearing that engineering skills is something that you look for. Are there any other types of skills that you sort of look for or any other soft skills like we mentioned earlier?

Zach: So we’ve talked about all of them here, but probably the biggest one is a willingness to work hard. That’s incredibly important here, especially since as a startup, we have a lot of ambition. We want people who are willing to work hard. I would say the second thing there is those soft skills we discussed. So being an effective communicator is unbelievably important. That actually might be more important than working hard. If you do a whole bunch of hard work, it doesn’t matter if you can’t communicate effectively. So those are kind of softer skills. The technical skills we look for tend to be more focused on engineering because we’re trying to build a product. And then I think honestly, even for people we’re hiring on the data science team at this point, we’re looking for engineers, even maybe more than we’re looking for people who are expert data scientists. If you’re a really great engineer and an intermediate level data scientist, you can build really cool stuff. If you’re an expert data scientist, but you don’t know any software engineering in Python, it’s a lot harder for you to go just build a new product on your own. And I think that ties into a lot of the frustrations you hear around data science is like, you need to be able to take what you’ve done and then do something with it. You need to be able to make a decision based off your model. And that takes business experience. You need to be able to build your model into a product. And that takes engineering experience. And I think there’s a lot of new data scientists out there. And if you want to differentiate yourself, having really good business knowledge or having really good engineering knowledge is a really excellent way of making yourself stand out.

Patrick Wong: Yeah, that’s really great advice. And I think that one thing that a lot of aspiring data scientists might consider is going to one of these data science boot camps. They’ve sort of exploded in popularity. But one thing that I constantly question is whether it’s actually valuable enough to put up the high cost of entering one of these programs. What’s your opinion on that?

Zach: Yeah, so I am somewhat skeptical of data science boot camps. I’ve got a friend who runs a data science boot camp. So to me, I see more value in engineering boot camps. And the reason I see value is there’s a lot of engineering skills that you can learn that, like we discussed, will be practical for many years to come, will be applicable in many different fields. The issue I have with data science boot camps is that you’re not learning a lot of those engineering skills. You’re focusing on the modeling process. And my advice to people, like I had a friend who wanted to get into data science, and he was considering a boot camp. And ultimately, the advice I gave him, which I felt was good and I think has worked out for him, was to try to learn data science on his own outside of work, and then to pull that knowledge into his current job in a way that was useful. And I think that in a lot of cases, that will lead to a better outcome than a data science boot camp. Because one, you keep your job and you can pay the rent while you’re doing it. But then two, it also focuses your education on where I think the biggest problem is, which is like, how do I apply this to real problems? And that’s way more difficult. So starting in an environment where you already know what the problems are, and then trying to figure out how to apply data science to them, that’s going to set you up better for success. You can figure out, man, we’ve really got this pain point with targeting customers for whatever. Maybe I could build a model that would help us target customers better. And since you have a lot of knowledge about what you’ve tried in the past and why it has or hasn’t worked, you’re going to be more successful there because you already work there. The other strategy that you can do is you can look for a company that’s doing really exciting data science stuff. And then look for job openings where you can apply your current skill set in accounting or project management or business analytics or whatever it is. And companies that are doing a lot of data science, if you can come in and contribute to real problems they have and get work done, you can also use that as a way to pull data science into your everyday work.

Patrick Wong: Yeah, that’s really awesome advice. I think that that’s sort of a strategy that I took when I was on the path to becoming a data scientist myself. So I think that finding a company that allows you to be able to practice data science is a really great pathway into the field.

Zach: Yeah, I would agree with that. You also mentioned Kaggle competitions as a way that you kind of fell into data science and got involved with it. So is that something that you would also recommend to aspiring data scientists?

Zach: Yes, I definitely would. So I think Kaggle’s got some problems. One of them is you’re usually given a really nice, clean data set to start with. And that’s never the starting point in the real world.

Patrick Wong: Never.

Zach: Yeah, But, I mean, I learned a lot of what I know about data science on Kaggle. There’s a really great community there. There’s a lot of good education to be had there. a friend of mine, a former co-worker of mine, Dan Becker, is teaching a bunch of data science classes on Kaggle now. He’s got some excellent deep learning classes. If that’s what you’re passionate about, he certainly is. And he’s a great teacher. So there’s a really good data science community on Kaggle. And it also, it gives you a chance to start to prove that, like, what you’re learning, you can apply to real world problems. Because even though the data sets you get tend to be kind of clean and you’re starting from a little bit of an artificial starting point, you’re still at least demonstrating that you can take some real data and solve a real problem. And then you can self-evaluate it on how your solution stacks up against, like, a few other thousand data scientists. And, like, that can be a really humbling experience. But it’s, it’s a good place to learn. The competitive aspect of it makes it fun. And so, like, it’s a good learning tool. Like, if you’ve taken some data camp classes, try your skills out on Kaggle and see how you stack up against the best in the world. And then, that starts to let like, do you know what you’re doing or not when it comes to data science?

Patrick Wong: So to sort of sum up your advice for any aspiring data scientists, learn a programming language like Python.

Zach: Yep.

Patrick Wong: Enroll in both your data camp classes.

Zach: Yes, Yes, please.

Patrick Wong: Try to find a company or try to find some work in your own company, your current company, to apply some of the things that you’ve been learning. And then if not, then you can try to find a company that does allow you to do that.

Zach: Yeah.

Patrick Wong: Other than that, it’s really just about making things happen on your own. So really just taking initiative and either building a solution to something or, building an entire product.

Zach: Yeah.

Patrick Wong: So I think that’s all really amazing advice. Thank you. Is there anything else that you’d like to share in regards to that?

Zach: Not really. I think, I’m just thinking about our conversation. One thing that you said really resonates with me when we’re discussing just how unglamorous a lot of data science jobs are when you actually look at their day to day. And that’s something I really want to emphasize. Data science isn’t going to be a magic path to career heaven. It can be really interesting work. In order to be successful, you have to be willing to work hard. And you’re not always going to be working on things that are the most fun possible thing you could be doing. You make time out of work to do fun things like Kaggle.

Patrick Wong: Absolutely. And I would definitely follow that advice. So I think we’re about out of time here. But I really want to thank you for joining us on the Accidental Engineer today. And I want to encourage anyone out there to check out your Data Camp courses. Your latest one is called Advanced Deep Learning with Keras and Python. And what was the other one called again?

Zach: The other one is called the Machine Learning Toolbox on Data Camp.

Patrick Wong: Okay. Got it. Awesome. So, Zach, thanks so much for being here again. This has been a really interesting conversation. So thanks so much.

Zach: Yeah, Thank you for having me.

Patrick Wong: For more, visit us on iTunes or our website at theaccidentalengineer.com