What We Discuss:

• What is Data Science?
• Data Scientist vs Data Analyst
• How To Improve In Interviews?
• Can you switch from Dev to Data Science?
• Do you have to be Good at Math?

Transcription:

what does that everyone welcome back to junior to senior developer. I’m Tyler Lemke.

[00:00:06] I’m Douglas Hirsch. And here, here, we’re here today with Eric Webber. I’ve got it right this time. Um, Um, he’s he’s currently the head of data science and insights at list reports. He started his career in academia where he spent five years, both teaching and researching, and he has been doing data sciency things since 2008 and has worked at other companies such as LinkedIn.

[00:00:28] And CoreLogic Eric. Thanks so much for taking some time tonight to meet with us. We’re really excited to talk to you. Well, thanks guys for the invite. Um, I’m excited for a couple of reasons. One is to just, I’m used to being on in conversations where I’m talking to data scientists and one, I think that gets all.

[00:00:47] Don’t tell them. I don’t tell him. I said that, but to, um, I think the important part is this field is hard to define and when other people hear it, when they hear the term data science it’s sometimes. Unclear what it is, what the job entails. Um, and I’m looking forward to getting to have that conversation along with other topics.

[00:01:07] I think that’s a, that’s probably a really good place to start at. So you’ve um, and one of, uh, one of an insert in a recent interview, you kind of mentioned a few of these different things. You said how, um, I don’t know if it was what the exact context was, but how companies want to like define what data science is to start off with.

[00:01:26] And, and they, um, uh, part of it is knowing the aspect of. What you’re going to teach within data science. So can you kind of break down the different categories of data science, because we all hear this word and it’s kind of a nebulous thing. We all have nebulous terms inside of the text. So it might be good to kind of separate out what you think are the main categories or aspects of data science.

[00:01:48] So I think I could answer that. I’m going to answer the question in a way that fits in a few minutes, but really this merits like many, many discussions because companies have these discussions all the time. Cause they don’t know how to define that. What it is, I’m at a high level. I, this is my definition.

[00:02:07] Right. And reasonable people disagree. I tend to think data science is all about the practice of getting value out of your data. Whatever that is real business value out of your data. And that is a pretty broad definition. And I think that’s okay to have that broad definition. Um, because often when you hear people define data science, they think like, Oh, they do machine learning or they build algorithms or they’re shipping something into production.

[00:02:35]Um, or they are working on what they don’t have a concept of is like how much basic scripting. And a SQL and Excel and things like this are involved in getting value out of data. Um, data science to me is more of an ecosystem, right? There are different roles that are a part of that ecosystem. Um, in some cases it’s people that make sure that they are getting the data prepared and ready in such a way that.

[00:03:06] People can actually use it. They can look at it. Right. Um, data engineering is a foundational, even more so than data science roles. Data engineering is probably growing faster and will continue to grow faster. As people need to have data to use, need to have it in some structured reasonable format in a timely manner.

[00:03:26]Um, Is it an analysis component of data, right? People that are doing this is where people tend to think about statistics. They tend to think about running tests on data. They run experiments to try to determine whether something is true, right? If you’re shipping a feature into a product, if you want someone to decide if that feature has meaningful impact on user engagement.

[00:03:50] That is an experiment or something that a data scientist might create. They might do the subsequent analysis after that as well. Um, there there’s another component to data science, which is, which I think tends to be what people envision is that we’re sitting here building novel algorithms and then putting those things into use and the product that’s actually relatively small part.

[00:04:13] I think of what a day is data sciences does. It’s. Well, when you start talking about novel development of algorithms, shipping those things into production, you’re talking much more about like in machine learning engineer type space, where. It’s not just the development of the algorithm, but it’s making sure that it runs in production in a timely manner, that it doesn’t negatively affect the user experience.

[00:04:39]Um, and so when it comes to like this whole different family of making, of getting value out of data, there’s a whole bunch of different job titles. And I think what you tend to see. As companies say data science, because they just don’t want to spend time giving you five different titles. Um, so this, you can imagine the confusion that is out there for people who are not working in this space on a daily basis is they’re like, well, I thought data science was this.

[00:05:08] Well, I thought it was this and there’s all this disagreement. And it produces these like. In my opinion, some fruitless arguments about what data science is. It’s more about like, are you, are you helping to get value out of data? Sure. What role you play that’s for you and the company to define. You know, I really like the way that you put that.

[00:05:29] Cause it felt like for a long time, even in just playing software development, we were just like this one, like software engineer was the job. Like no one, no one even differentiated us. Um, front end backend. In fact, it was only when I got to Texas that I actually learned when I told someone what I did.

[00:05:47] They’re like, Oh, you’re a full stack developer. That’s awesome. But apparently at some point I didn’t know about this. Um, what we did finally got split out into like front end, back end design and you know, and database development and all this stuff. And the way that I was raised up in the industry, Uh, basically we just did everything.

[00:06:08] You had to learn how databases work. We had to write our queries and get value from that data, but then also know how to store the value in a way that we could get value from that data. Just like we were talking about, um, all the way through making our front ends with a, you know, HTML, CSS and Java script and back ends with whatever we were going to use.

[00:06:26] And, and, um, Yeah, it was all over the place. Cause no one knew how to separate that out. So I think from what you’re telling me is that at some point they’re going to get smarter about this. They’re not going to call it data science. It be more adult they’ll have the different aspects. Cause it’s a much younger, it’s a much younger designation, I think, you know, software development.

[00:06:48] So this makes a lot of sense. I think they’re gonna, I would hope they get smarter about it. Right? That’s the best case scenario. The more likely scenario is. That just takes a long time for these differentiations to really take hold. Um, you’re starting to see these differentiations happen much more at larger companies.

[00:07:08]Um, ones who have been trying to work in this space for quite some time. You’ll see a lot of major reorg. Sort of happened at Google and Facebook and LinkedIn and those types of companies, precisely because they’ve come across the same issue, they’ve built large data science departments, and then they figure out that, well, actually we’re doing like six different things and we all shouldn’t be part of the same organization.

[00:07:33] Some of us should be part of marketing and sales. Some of us should be part of engineering. Some of us should be part of a full fledged data organization and. What I’m seeing is until a, company’s have a reason to not call it data science. They tend to just do it out of ease of use, like smaller companies, especially there’s like, yeah, you work with data.

[00:07:57] We’ll just call you a data scientist. But it produces this, um, where it does produce a lot of issues in companies as they’re trying to hire people because. If you go off titles titles, in my opinion, in this space are really hard to assign value to, but if you have a recruiter looking at a bunch of resumes and they look and see, like someone was a data scientist somewhere else that leads them to believe they have relevant experience and it’s going to be right for this role.

[00:08:29] And that’s not necessarily true also because data science, in my opinion, when you think about getting value out of data, It’s very dependent on being an advanced to expert in the context of the company’s operating. And it takes you awhile to learn if you’re doing data in real estate or you’re doing data in healthcare, or you’re working for this particular company.

[00:08:55] A lot of the skills might seem translatable, but they’re often very context dependent and that takes a long time to ramp somebody up. Like, to me, you’re not really effective until you’re like six to nine months into your job probably. That makes sense. I mean the domain, right. Domain, whatever domain you’re in.

[00:09:13] Right. It’s it’s um, it happens. I’ve been in, uh, several different domains through, through my career. And I’ve noticed that, that you really can’t have business conversations with people and know how to translate what they’re saying into code until you actually understand the business side of what’s what’s happening.

[00:09:30] It’s extremely important. So I. I know where you’re coming from on that. It takes a while to get someone up and they have to want to get up to speed on that. But I think on, I think in the data science world, if you, if you’re trying to get a job in that realm, and that is your job, you’re going to want to get up to speed as quick as possible.

[00:09:48] And the domains that you can have, those domain specific conversations. Absolutely. So you, so I think this is a good, a good segue into, I think something that you talked about before, um, is with the domain knowledge, um, another interview you had a couple of years ago, uh, there was a question that came up about mastering the data.

[00:10:10] Does that sound like a familiar term to you at all? Yeah. I’m always like people, it’s sometimes funny people, like you said this a couple of years ago. I’m like, I hope my previous version of myself said something so many intelligent. And if he didn’t then Nope, I don’t remember at all. But yes, mastering the data.

[00:10:27] It’s it’s, it’s something I would have described cause I continue to believe it’s important. So, what does that, so you talked about how, where the data lives, how to access the data. Can you talk a little bit more about this and how can a junior or mid level data scientist kind of get up to speed faster with mastering the data and whatever role they’re in or where they’re at?

[00:10:52] I think to me, You don’t learn about you don’t master data in the abstract. Like no one very few companies is someone going to present you with a data model. And from what you’re going to be able to infer how to, um, join different data sets together, how you’re going to be able to leverage them and affect and, and, you know, to create effective insights, you really have to be tossed into the deep end with a problem.

[00:11:22] And so to me, like when people talk about onboarding a data scientist and that you kind of make or break somebody within the first six months, if they’re going to stay with you for awhile, um, you have to give them the right problem to dive into when you start. And that problem should be broad enough that it forces them to.

[00:11:41] Look at more than one single database or one single table within a database. Um, it should require them to particularly in like a setting where you have customers and user behavior looking at user behavior across different contexts, time periods, stuff like that. Because what we, the reason that I think it’s so valuable is until you learn, um, The ins and outs of the data that you have.

[00:12:09] It’s hard for you to know what the limits of the questions you can ask are, right? Like if, if a business partner comes to you and they’re like, Hey, like I’m wondering about this. I’m not sure how to figure it out. Like this is confusing for me. It’s really important that you’re able to. Now, if it’s possible to answer that question, because.

[00:12:29] It definitely is dangerous if you’re like, yep. I’ll work on it. And then you’re like, Oh, sorry, I couldn’t do it. Like we don’t have the data. Um, or if you don’t have the data and you need to, to be able to call that out, right? Like we’re, we need to be collecting this. We have no insight into what the user is doing.

[00:12:46] Right. There’s a lot of companies I’ve joined where they’re like, we’re trying to map the user experience from like, from the moment of sign up until 30 minutes after sign up. And then I tell them that, well, you don’t have any actual events that you’re tracking between those points in time. So your ability to decide, describe anything is not going to be good.

[00:13:09]Um, And so it’s that onboarding experience it’s like really feeling like what are your tools? Because people like to talk about their tools like it’s R or Python or SQL or whatever it might be, but your tool usually is knowing the data super well. Um, that to me is. More important than whatever code you are writing is you have to start from somewhere like, and I think as you see more in the, we’ll get into this topic later, but you see more senior people, more senior people tend to really, when they come into a new company, that’s what they push.

[00:13:45] Hard-on right away. I’m like, where’s the data. How do I get to it? Can I use it? Can I bring it together? Um, and so that’s one of the major differentiators that I see just in terms of maturity in the space. Okay. So it sounds like if it’s really, it’s something that’s really difficult to do without some mentorship from your peer or from your, uh, you know, higher ups or peer like peers being like more senior or your team lead or whoever it is.

[00:14:14] I don’t know how it works with data science, but if you guys have team leads typically, or how does that work? I think it depends on the thing that’s most frustrating about this space is. The answer. It depends seems to be the most common answer, right? It’s like depends on the size of the company and the maturity of the company.

[00:14:33]Um, in an, I I’ll talk from like an ideal scenario and an ideal scenario, you have a pretty intense onboarding experience that lasts a couple of weeks with someone who is not just a business partner or a business expert, but who is a technical lead. Who can teach you how to get to things and why you need to get to things at a certain time, um, why it’s important that you access this database and this way they can, they can often help you overcome many hurdles that you don’t even know are there by, and, and they can do it really fast.

[00:15:13]Um, what I often see though in the onboarding experience is like, Okay. You have, you know, here’s your computer. We give you your log in info. Godspeed. That learning experience is not pleasant for anybody. Um, and so I think about that a lot. And part of that is like,

[00:15:39] The field is new enough that like what mentoring means, and that is still like, I think being defined a bit, it’s hard to actually find really good mentors because many of us are still figuring it out. And I think that’s a massive challenge, honestly, for not only starting a company with a couple of data scientists, but if you want to scale a team, you have to have people to mentor that team.

[00:16:03] Not everybody’s a good mentor in order. They want to be one.

[00:16:09] let’s just say that, that, that kind of rings true. What you just said in the software field too, because depending on the size of the company, you may or may not. And this is what we tell. I actually, uh, I’m an instructor at a coding boot camp right now, and I’ve been in field for a long time, as far as being a software developer and.

[00:16:27] What I tell my students is that we really want you to go into a company that has a structure around mentorship and getting you up to speed, because I’ve actually had developers that outside of coding, boot camps, that I’ve mentored where they got a job and there’s no way they should have been put in that position.

[00:16:45] They were put in. There were no other developers and yeah. They were told Godspeed at junior level is, did not end with a success story. It never does. I actually had to go and I talk them off the ledge basically. I’m like, dude, you’re fine. This was their fault. You, you look, this is not how I would have done this.

[00:17:02] And I’m sorry I introduced this vintage of this company. So I thought that it was, they were at a different place. And then he told me what happened. I’m like, no, man, I’m really sorry, but you. You’re fine. The company the company did to you was not. So we see this, we see this in the software development field too, and I wish to say that it gets better.

[00:17:22] But I think that, that we have a lot of organizations run by run by humans and they don’t necessarily always, they’re not always the way that we want them, but optimally speaking, you would always want that, that pipeline and effect right. Where you have your senior people and then more junior people come in and there’s that mentorship.

[00:17:41] There’s always a pipeline of people coming in because unfortunately in our field and I got a feeling it’s the way in data science, too. There’s always people leaving as well as coming in. Always, you always want to have a good, you always want to have a good pipeline and you have to be a certain size company before you can actually have a pipeline of people coming in.

[00:18:00] Yeah, it’s a, I mean, it’s a huge problem. One is you have people leaving on a regular interval, right? In any, in any space like this, where. Companies are, you know, like right at this moment where certain companies are being super aggressive about going out and hiring top talent, you’re having people leave.

[00:18:28] And I think whether times are good or times are bad, it’s still going to be true. Um, and so if all of the onboarding or all of the mentoring was the responsibility of this one person who was done suddenly gone. Like, and the people that were at that person’s level are not interested in mentoring, like your mentoring program dies and the onboarding experience dies like, okay.

[00:18:56] I think to me and I actually, you know, in a perfect world, perfect world actually existed. Mmm. I would think seriously about science. Like if there’s somebody that you expect to be a mentor, And a manager. I literally try to sign them to some form of an incentivized contract to keep them around longer with escalators, with like, because what I see happen so often is okay, an individual contributor, the loss hurts, but if they were solely doing technical work, You have at least a shot at replacing them.

[00:19:34] That’s not a guarantee. You have a shop that you might replace them. If you lose a good leader who is also like, it’s so hard to get that team back in shape, it kind of throws things into disarray. Um, particularly in this space, like it puts the junior people at risk. And like, if you, like, I just I’ve seen it happen many times.

[00:19:58] Where suddenly the junior people came in with the expectation, they would be mentored by this person, that person leaves. And then the company has a whole mess on their hands because that team essentially dissolves. And if I were running a company, I’d really think carefully, like I understand competitive competitiveness that people are always going to have opportunities to leave, but I would think really hard about building a combination of culture and incentives that are going to keep people.

[00:20:27] In place for at least a period of time. Right. So they’re not bouncing out after six months. I think you make, I think you make a great point there and to bring it back to people that are listening. Cause that’s, I think that’s a great thing for ’em. Any any, uh, uh, employers that are taught, uh, might be listening.

[00:20:46] So I had a follow up question to that though. How can you, how can you as an interview E write a, um, uh, how can you identify? Mmm. If someone on your team is going to be a good mentor, when you step into that role so that you can set yourself up for success as much as possible. Yeah. That’s a good question.

[00:21:07]Um, As an interviewee, I tend to look specifically at, have they done it before? Have they done it before? Will they allow me to chat with someone who’s actually been mentored by this person before? Like I realized that sounds like an idealistic scenario, but I think from an interviewee’s standpoint, like it’s worth the investment of time to.

[00:21:35] Ask us person, right? If you interview with them, you have the five to 10 minutes of the interview, your interview, ask them how they think about mentoring, right? Like most people are gonna, you’ll be able to tell if they’ve at least thought about the question, right? If I give you some answer, that’s like, aha, that’s not accurate at all.

[00:21:58] It doesn’t work for me. It’s good. It tells you something, um, Most likely interview loop. You’re also going to probably be interviewing with someone who has been mentored by that person, right? Whether it’s a manager, if you’re talking to the tech lead of the manager, there’s very likely someone else in your interview loop that works on that team for that person.

[00:22:20] And if there’s not, then it’s also reasonable. I think particularly if you get to a final stage where they’re trying to close on you. Ask to speak with somebody that has been mentored or has experience with that person. Of course, there’s a lot of gray area and the information you’re going to get, but ask people to speak to their experience.

[00:22:45]Um, I find that it’s really illuminating. It’s often, like it’s illuminating in what people tell you. And then also what they don’t tell you, if you ask, like, what’s the best experience you’ve had working with this person and they tell you about something that doesn’t sound fun at all, that tells you a whole lot about what it’s like to work at that person.

[00:23:05] Right? I’ve often found that people, mentors who are amazing people will just go to bat for them. People can not say enough, good, positive things about somebody. If they really enjoy being mentored by that person, you just got to ask questions. Um, it’s hard, but it’s hard. It’s hard. And it’s also kind of scary, right.

[00:23:25] Particularly at a junior level. Like I’m just trying to find a job, but I, I tend to, when I work with people who are coming into this field, I tend to have them think and like, okay, I think about the potential, like we’re really bad as humans at evaluating risks. You know, COVID is a good example. Like we’re really bad at evaluating risk.

[00:23:47] When we go into thinking about a job we’re also bad at evaluating risk. Like what’s the risk. If I get into this position and it’s not good, and something goes off the rails, how does that negatively affect me later? The amount of work it takes to turn that around the amount of drain on your mental health, like.

[00:24:08] All of that is present, but I think people are not usually thinking about the risks of a role when they’re looking to join. They’re thinking like, alright, I need to get a job. So I find that to be a tough balance. Um, yeah. Oh, that was an amazing answer. I hope everyone just like goes back and like plays back a couple of last couple of minutes and like takes a bunch of notes.

[00:24:28] Cause that was awesome. What do they say? Sometimes I the same things. And I’m like, what was that again? Yeah. I don’t remember, but my thinking, my thought stream during all that was like, this is amazing. Okay. That’s funny. You’re like asking me, like, sometimes I’ll be giving presentations and then someone will ask me a question about something I said, and I’m like, What was that said by me, I’m not sure where I was going with that.

[00:24:58] What word? Okay. Hold on. Let me figure out an answer, man. So, so no, it’s a, I think it’s really cool that you talk about asking questions. Cause I think people don’t ask enough questions and that’s, that’s where, um, I mean, even down to, and I don’t know if any of my students will ever listen to this or not, but if they do, you should ask a lot more questions of us.

[00:25:21] That’s one of the things you’ve taught too. So you, you kind of know. You kind of notice, it’s like, Hey, I can almost answer anything you throw at me. And if I can’t, I can find the answer. So throw stuff at me. And then you say, Hey, see when you need help with this. And they’re like, no, we’re good. But she does not.

[00:25:40]Um, yeah, usually it’s like the person that’s brave enough to ask the questions tends to be a good signal about what kind of. What it’s like to work with them, like in an interview setting. Like there’s nothing that I’m more than I enjoy them when someone’s asking me questions where I’m just not the one firing questions directly at them.

[00:26:00] Like if someone has the confidence and the belief in themselves to interrupt me and ask a question and say like, Hey, I’m not sure. Like, what do you mean by this? Like, and the last 10 minutes of the interview, contrary to popular belief, it’s not just a formality that they give you time to ask questions.

[00:26:17] Right? It’s a. Literal chance to show that you’ve done your homework on a company, on a person, um, that you are looking out for yourself as much as they are trying to look out for themselves. It’s such a valuable five or 10 minutes that you have, but when people think about interview prep, it all goes into like how many questions that I prep for?

[00:26:41] How many, like it’s very rarely about, do I have a good list of questions? At least that’s been my own experience. So, I mean, that’s, that’s kind of where it is for a long time. I think as you get more and more senior and you know that you’re okay with the technical questions that you start getting into, um, How, how has your business run?

[00:27:01] Exactly. You know, how, how do you handle this? And that, like the last time, my last interview, I actually asked more questions and they asked of me, it actually turned into, it, turned into that as a kind of I’d had experience at another coding bootcamp. And so I actually knew all the things could go wrong.

[00:27:18] And so I started asking them, how are you handling this? How are you emailing that? And that Sergeant tool way different interview experience than having people throw. Questions at me and try to me trying to convince them that I know where they were trying to convince me that they, they were a good place for me to come work at it.

[00:27:34] And that’s, we always try to, we always try to tell our students that, that the interview is bi-directional. You need to ask questions. So you feel comfortable. It’s not about just getting your first job, although, and that’s how, tell me what you think about that. Cause right now we’re in a very different situation where competitiveness has changed a lot.

[00:27:55] So the asking questions thing and getting. Maybe what someone might consider a little more aggressive about asking questions. There may be a little fear there about letting him, I really need this job. So I want to stay quiet and just answer their questions so I can, you know, so I can get a job. But what are your thoughts on that?

[00:28:14] I think I mentioned preparing for questions for an interview that doesn’t just mean, uh, Creating a list of questions. It actually means to me asking the questions that somebody else and having them give you feedback on the tone of your question, how you asked it. How you followed up on those questions is typically, like, if you think about most of communication, it’s very rarely also you want someone who’s going to call you out and be like, that question is not something you should ask a prospective employer.

[00:28:51] Like I think I posted the other day, I’m like someone asked me, they’re like, Hey, I saw on your Facebook that you have a brother and a sister, like, don’t ask that it’s a bad idea. Like, I was like, really? Like, what else have you been watching me? Um, but in general, You should really make sure it like dry run your questions and think about the tone that you used to ask them, make sure that the wording is right.

[00:29:14] Make sure that it is relevant for the company. Um, so much of it is in the body language that you use to ask the question, right? People like hate this clinical communication, blah, blah, blah. But I can, it matters a lot. How you ask the question, how you present yourself during it, like practice sitting there, practice being confident like that.

[00:29:38] Like people feel like they’re in speech class again, and I’m like, I can tell you like the difference in impact versus where you were like, have. Thought through the question ahead of time. And you’re confident in what you’re asking versus someone who is like slouched in their seat, kind of talking near the floor total it’s day, night and day.

[00:29:56] And it’s this weird thing, but I think so many people in all of our fields, like engineering, data, science, whatever it may be. Thinking a lot about how you’re communicating and people think about that verbally, but like your nonverbal communication is super important during an interview. Um, particularly for junior levels.

[00:30:18] So the, uh, the other thing, we, we actually talk about this a lot, cause, um, we, we produced a good number of junior developers and, uh, interestingly enough, you might find this a bit interesting. We do have a data science program and you’re like, Oh man. But no, we, we do have a, we actually do have a data science program and we have some really good people in our data science program and some very good hiring that are ready to take on.

[00:30:41] For students. And, um, um, and so the thing that we always talk about is how to ask a good question. Cause you don’t want to be that that asks, uh, it asks us to be the stupidest questions. I, I hate to use that term because we always say there’s no such thing as a stupid, the only stupid question is the question not asked.

[00:31:00] Right. But there, there are times where you have, uh, You’re overstepping or you’re really, you know, that senior is trying to get something done and you could have answered that question yourself. What do you think is a, uh, uh, you have like a good set of how to ask good questions. So instead of, you know, it’s really funny, like my literal last, uh, Last LinkedIn, LinkedIn posts that I made like an hour ago, it was like ever gotten to the end of interview.

[00:31:29] And like, here’s some questions that I like to ask. Like, ironically, this is literally what I posted one hour ago. Um, so when I think about these questions, right, like what they are, why, how I, how I assess whether they’re going to be considered valuable by the employer. It’s it’s hard, but like when someone’s being inquisitive, like you want to show curiosity, you want to show curiosity in that you’re curious about what the company is doing, where they’re headed next.

[00:32:00] Like, how do you fit in to their overall vision? Because what people want when they’re deciding whether to hire you beyond technical skills and they want to feel like they can come and work with you day to day. So asking questions about how you work in a team, right? Like how do we get things done together?

[00:32:22] If there’s a big project, like how do we share responsibilities on it? If someone is sick, how do we cover for that person to make sure that things continue to go out the door at a reasonable pace? Like. I often think when you show that your questions are about you wanting to contribute and be a part of a team, anything within that domain communicates one that you are excited about the opportunity to that you think of yourself as part of a team.

[00:32:54] And almost any question within that domain is fair game. Right. It’s like, okay. So how do we work best? What has been your experience? And like, you know, mentorship falls into that, um, working as a team, how do you think about code reviews with each other? Like, how does that work? How do you make sure that when like that we make, so let’s say when a director does a review of a team, how do you make sure that.

[00:33:22] The team has done collective work that makes the manager and the team look good. Like, you know, that they’ve met their goals for that quarter. I often find that resonates with people because the communicates that you think of yourself as not just an individual, um, and that’s what they want. They want to, and also like, It’s it’s so hard to quantify.

[00:33:45] I think this is the interesting part of our interview is we want to quantify so many things, but like people just want to know that they would have a good partner to work alongside someone who could do their job. It could also be fun to work around. And that seems so basic, but it’s actually pretty complex.

[00:34:03] So ask questions that show interest that show something about your personality, right? Of course. Keep a gate on that. Don’t go to like often to left field about what questions you’re asking, but, and this again is like, if you, before you ever go interview, try these questions out in somebody else. Like if that’s the one piece of advice that you give people, ask someone else’s question and get, get their feedback.

[00:34:31] If you happen to know a hiring manager, not necessarily like a friend of yours, it’s a hiring manager. It might be really good to run that set of questions past them, just to be sure that it doesn’t, uh, it doesn’t rub them the wrong way, because that might actually be a good, I love this. I bet you just did it again.

[00:34:48] I know the past couple of minutes, we, they, people need to listen to it a couple of times, because what you just said is we’ve got some gold in here. Yeah, I’m putting that in the show notes. Actually it comes from experience. Like, I think where people, I mean, this is also something that when you’re trying to enter into this area, whatever it might be like, you should use each experience to try to learn something from like, if you had a bad interview, don’t just like block it out, figure out like what went wrong.

[00:35:22]Um, Sometimes the recruiter will give you feedback sometimes not, but like after I interview, I usually go out to my car and I like do a voice recording of what I thought about the interview, um, for like 10 minutes, mostly because that’s the only time I’m ever going to remember what actually happened that day, because then you get tired.

[00:35:41]Um, but it’s actually interesting to go back and listen to like the recording I did after my interview at LinkedIn. Versus after my interview at some other places that just didn’t go as well. I pretty much knew what happened at that, like right after, but I think what’s different is like a day removed.

[00:36:05] Be like, okay, that went well or that didn’t. But if you do it immediately after you often know exactly the details about what went well and what didn’t, if they later, you’re not going to remember what, what happened. I love that idea of doing like a postmortem and trying to figure out what, what, what went well or what went wrong and recording out.

[00:36:24] It’s really great. And cool thing about it. Well, Oh, I was gonna say the cool thing about that. And I didn’t mean to, I didn’t mean to interrupt you, Tyler, but. I wanted to throw this in there. The really cool thing about doing that, that post-mortem right then is that you can use that as like a hypothesis as to the result of your interview and have recorded that and then get the results later, which is really cool and learning, uh, how you know, learning about even your ability to estimate.

[00:36:55] How, you know, to hypothesize about how well something went. Um, I, that’s amazing. I never really thought about doing it right then. There’s all these memories I’ve got over the years of interviews in my head and I don’t, you know, I know how we, I, you know, we rewrite memories as we bring them out. So it’s a think about them.

[00:37:15] And so I look back down, it’s been so cool. If I’d had just a five minute. Five minute video or recording about these interviews that I remember that went so terribly wrong. I know, just because I say record doesn’t mean it’s a positive thing sometimes it’s like, you can tell that you were traumatized.

[00:37:37] Like I didn’t. I remember the first time that I, I went into this was with, with Amazon years ago and it was. At the time where they were still figuring out their fields of data scientist, applied scientists, research, scientists, differentiated the job title. Well applied scientists. How elite code hards are like a norm and I was not ready in any way, shape or form.

[00:38:02] Right. So I walked in and they’re like, we’re doing like these things that I now recognize as like weak code, hard questions at the moment. I just wanted to cry my interview tape after that was like, this is the worst day ever. I don’t know anything, but at the same time, there’s really positive experiences too.

[00:38:20] And like, it can be really informative. Just from like a, you know, I’ve probably done. I think probably in my life, I’ve been probably like 40 to 45 onsite interviews now. And I have recordings for at least like 35 of them. And it’s weird. And like, but you learn from it. Um, I just hope I don’t lose my phone I’m I started in the cloud now, so I’ll have them forever.

[00:38:46] They got him. I got to make sure to back them up. Um, so I think, uh, I think I want to say to another, another really important question, and I think it’s one I’ve kind of had before, or at least I’ve thought about is how can you transition as a, as a junior or mid level software developer into. Data science or data analytics or whatever one of these titles is.

[00:39:06] Right. How do you, um, so how can you do that? And, and what are the, what are the, um, character, character traits that might make someone, um, uh, a good, uh, candidate for transitioning over from software web development, into data science or data analytics? I think that’s a good question. Um,

[00:39:23]I’m just going to go off of experiences that I’ve actually seen. Like there’s what I think. And there’s what I’ve actually seen. I think they’re a little bit different. Um, I’ve found, I’ve seen it be very challenging for people to go entirely from like software dev into like an analytics role. Um, and by that, I mean, where they’re literally just running, they’re doing basically deep data investigations.

[00:39:47] From coming from a software dev side. I don’t know why. Maybe it’s just like my own bias and who I’ve interacted with, but I’ve seen far fewer people be interested and successful in that specific transition. What I typically see is software devs are thinking. About issues of scalability, they’re thinking about issues of user experience.

[00:40:12] And so when you think about one of the most common transitions right now, is people going into, um, ML engineer roles, because what’s what that is all about. It’s less about the analytics. It’s less about the data analysis. It’s more about figuring out how you deploy a complex Oliver algorithm at scale.

[00:40:35] Without having a bring down the user experience, right? How do you ship something at scale? How do you put it in? Um, how do you serve it up? How do you create recommendations in a way that is going to not disrupt the user experience? And so, because the most successful ML engineers, if you look at most roles that I’ve seen, they come from software engineering, that’s much more rare.

[00:41:00] At least from what I’ve observed again, this is my own bias. I play as much rare for a data scientist to go into an ML engineer role than it is for a software dev to go into an ML engineer role. Mostly because I think the way that software engineers tend to think yeah. Is really well suited to that type of transition.

[00:41:23] Okay. If you’re thinking about going wholesale, like I want to get into data science, like to me, The thing that defines someone, making that transition is really like, do you understand statistics and data visualization? Like those two things are paramount when it comes to that transition. Um, if you don’t enjoy statistics and data visualization and like a typical data science role is probably not going to be something that.

[00:41:55] You’re going to enjoy, um, if you still want to focus on our lives and deployments, like a more heavy engineering side can more heavy engineering role can be really valuable. That’s another thing I was going to ask because, um, today you don’t have to really be great at math to be a software developer software engineer, depending on where you’re working.

[00:42:15] Right. Or web developer. Um, like I’m not very good at math. So would you say like, would you have to be really good at math? Or could you just be like math being like, you have a really. Like, you know, you’ve gone into like discrete math and all this other different stuff. Or do you just have to give, be good at statistics and individualization?

[00:42:33] Can you just be good at a subsection of math? Cause you’re, you’re an expert. Like that’s my buddy. I mean, my expertise is in like theory and probability and stats, right? Like, and the mathematics behind it, but you don’t need that. Like, you don’t need that to be successful in these roles. You need to be able.

[00:42:53] Like quantitative reasoning. Like you need to be able to think creatively about problems. You need to like, if you were like a tenacious problem solver, math is a one component of that, but it’s rarely the only thing it’s like, I would say that math doesn’t differentiate. Good from great or great from good math is just something like, you’re probably not going to use math.

[00:43:19] That’s extraordinarily complex, almost ever. Um, when it gets into more advanced role, like statistics, like statistics, people think it’s like all about numbers. Statistics is more about how you interpret the numbers, how you create an experiment and then interpret the results in a way that’s meaningful.

[00:43:38] The math is typically abstracted away and the software at this point, like the actual calculations, the number of times I’ve done like a hand calculation or some mathematical thing by hand is almost zero. Like you start to think, like you offload a lot of the deep mathematics too. Um, statistical programs.

[00:44:02] And so that’s why, where I see a lot of value in things like learning how to do an R or in Python, um, because you can offload a lot of the heavier mathematics to focus on the interpretation and it’s that interpretation and understanding of the context that you’re in, that typically makes or break someone’s success in this field.

[00:44:23] So is this true with both machine learning and with just data science in general and, okay, so this is another good conversation, I think, to be excellent. And like, if you’re really going to be focusing on machine learning, um, there are two camps here. There are a lot of people who just apply machine learning algorithms and look at the results of applying those algorithms.

[00:44:51] That to me is how do I describe it? It’s not an ideal situation because what’s happening is that, I mean, if you think about it, they’re effectively just picking 10 tools up and they’re hitting the same problem with this tool 10 different times. And they’re like, which one of these produces the most positive result?

[00:45:13] When most of the interesting stuff is actually in why it produces. The result that it does, right? Like why did this technique perform better than this one? Why like, and over time you hone your skills so that, you know what technique is probably most apt for that problem. That’s where you start to get into more in mathematics where you start to really think about, um, Like were actual calculus and actual like rates of change and how you search for optimal solutions actually comes into play.

[00:45:45] But to be quite honest, with most data science roles right now come, and this is something that I actually considered to be a negative. The algorithms are more or less as something people. Click run on rather than something they think about carefully designing. And that actually produces a lot of P a lot of algorithms being built and a lot less understanding of why they’re doing what they’re doing.

[00:46:12] So think about what could happen if like, so a good example right now is. With, um, basically with COVID happening a lot of productional production machine learning algorithms that basically look at recency of user behavior to generate recommendations and predictions. They all just broke because people’s behavior completely changed.

[00:46:37] And so they were, the models were more longer term tuned on user behavior. That made sense. In this time or in this moment, but for someone who’s just building an algorithm and just applying the algorithm, like the mathematics of why that happened, wouldn’t make sense. And so to me, like this field is so deep, there’s so much here.

[00:47:02] That it’s more about figuring out where you’re going to dip your toe in the water, then like trying to do everything at once. Like you’re not going to be good at everything you need to start somewhere. And like that is, I think there’s different entry points into that. I think that’s, I think he touched on it and I might be speaking over, uh, Doug.

[00:47:21] Cause I know he has something he wants to touch on, but I think what you just touched on is really interesting. Cause I think that that is probably one of the keys of moving into setting up your career is, is going deep. And how, how do these things work versus, you know, can I use them. Right. Like understanding the algorithms behind them.

[00:47:39] If you want to go into machine learning and stuff like that. If I’m using the right terminology, I’m probably looking like getting idiot right now. Um, I’ll, uh, I’ll defer to, uh, uh, Doug as we wrap up here. Cool. So, no, I was gonna, I was gonna say that to our developer audience, uh, Eric, what you just said makes a lot of sense, even in our field too, because what happens is, as we’re learning things sometimes, and I’ve watched junior developers come in and they’ll poppy some, uh, some code off of stack overflow and paste it into their program.

[00:48:10] And then, then they’ll come get me and say, Hey. And this is not even a student. This is actually, this happened to me at a company that I believe it, it just kind of comes to kids and says, Hey, this Coast’s not working. Why is it not working? Stack overflow says it’s supposed to work. And I’m like, um, okay.

[00:48:27] So first of all, let’s take a look at the code and scan it line by line. So. Uh, you know, it, it, it turned out it like, like he took this wholesale copy and pasted it and expected it to even with the variables and everything that they had in there. And it’s like, Oh, this is supposed to give you an idea of how to solve the problem.

[00:48:44] This isn’t the actual coach who used to solve the problem. And it kind of reminded me when you said, when she said about, about people just using the algorithms without understanding it. Stack overflow could be its own discussion of a show, man. That is a, but yeah, it’s a, go ahead. We’ll just say we’re going to have one, one day.

[00:49:06] I don’t know. I mean, it’s tough because in a lot of ways that’s how people, like there’s a lot of informal learning routes into this field and there’s no like. There’s very few people that are like, these are best practices. Like people are kind of hacking and figuring out how to get there. And like this seems to work for this person.

[00:49:28] So I’m going to fork this person’s repo and get hub and like, I’m going to use their code and look, and then suddenly you permeate all of these really bad practices and they’ve spread across and, and it gets into, and so I really take seriously, like, The idea of accrediting, um, bootcamps and places that are training people, because I think there’s a lot of, um, it’s tough.

[00:49:59] We’re turning out people in some cases with the right skills. In some cases they’re being promised something. That’s not going to get them to the right spot. That’s frustrating from a personal level because they’ve invested a lot of time and money and then they like, Oh wait, I’m not actually ready. I know exactly where you’re coming from.

[00:50:16] The, uh, the one thing I will say about where I went to go work is that they really had, uh, I was on the fence about going back into working at a coding boot camp. And, uh, they really had to convince me they had a good program and I still firmly believe that they do, um, where I went to go work. So, uh, we, we even take, uh, we’re able to take VA money in there.

[00:50:35] They’re in the process of accreditation. Even now they’re working on trying to get to that point of actually being able to have an EDU domain and stuff like that. So they really take it seriously. It’s a really good program, but you’re right. It’s it is so much all over the place that, that as far as coding boot camps are concerned, we can, we could do a show on coding boot camps.

[00:50:57] Yes. Oh, that’s a whole conversation in itself.

[00:51:03] Yeah. So Eric, it’s been a, a word. Uh, I know we’re short on time here. It’s been a pleasure to talk to you so far was, uh, we got a few minutes left. Is there anything else that we should have asked you that we haven’t asked you yet? I’d like to ask that. I think it’s a, I think what I’ve learned in this space is anyone who says anything definitive about what data science is or isn’t.

[00:51:29] Is just trying to, like, they’re just making a statement in the moment. I actually don’t think it’s super helpful to try to make those determinations right now because companies are still learning. They’re trying to figure out what’s going on. They’re trying to figure out what the lines are like as data science and engineering practice.

[00:51:47] Is it its own practice. I don’t necessarily know. And so I think people want this want to accelerate its development when it’s actually okay. It’s okay. That it’s developing and weird ways that don’t make sense. And the titling is messed up because companies aren’t going to change unless they see something that’s not working for them.

[00:52:11] And that takes time. Like, but you know, we’re not very good at patients in general. I’m not going to patients. I’m like, we needed to find this needs to be perfect, but realistically it’s not necessarily a bad thing. It’s just like, it’s the development of any field. It takes time. And what data sciences 10 years from now is going to look a lot different than what it is now.

[00:52:33] And that’s okay. We just have to wait and be actively thinking about those things. Or we’re all having, we’re all having growing pains. Right. All men is only about seven years old and you know, other fields are a lot, a lot older, you know? So, uh, it’s, it’s interesting to see, and it’s been really cool to get your perspective on this stuff, cause I’m definitely, I know know very little about data science.

[00:52:57] So been really neat to, to hear about that. What’s the best way people can connect with you. Eric, I’d have to say like LinkedIn is my centralized place. Find me there. Um, send me a message. I can’t promise that I return it that same day. Um, but I try to be really active, um, in responding and making sure that we take care of, um, That I, you know, get back to you as much as I can provide what insight I can.

[00:53:22]Um, I tend to post pretty frequently, especially during lockdown because I find my mind is just like spinning and I have all these ideas. So interact with interactively there. Also, if you see me posting like. Post questions or responses. Like you’d be amazed at how many people in this community want to talk about things and they just want to have good discussions, um, jump in and you’ll see a lot of really interesting feedback.

[00:53:48] You got some off the hook posts of like thousands and thousands of a response every now and then it’s really weird. I’m like sometimes I wake up and I’m like, Oh, I shouldn’t have said that because I literally try to respond to every comment. And sometimes I wake up and realize that. My morning is going to be shot to like, try to respond.

[00:54:09] But it’s fun. It’s also this really good experience. Cause I think it’s a two way street, right? You’re going to have engage people when you show that you’re engaged and are willing to learn from them. Awesome. Well, thanks so much again, Eric and we’ll we’ll uh, we’ll be connecting with you and commenting on LinkedIn too.

[00:54:27] All sounds good. And thanks guys.