comet-april-1.mp3 Speaker1: [00:00:09] Welcome to the comet Emelle office hours powered by the @TheArtistsOfDataScience, super excited to have all of you guys here. Thank you so much for taking time out of your Sunday to join us today. Hopefully, I've got a chance to tune in to the art of their Science podcast episode on Friday, where I talked to one of my favorite people, John. I had a call from LinkedIn is a great conversation. Also, don't forget next week to catch IOW myself at the Data Science Virtual Conference, we're going to be hosting a panel discussion. And it was actually just recently announced that I'll be emceeing the entire thing. So the heads up next week, I will likely not be here for office hours. Sorry, Odelia. Dropping the bomb on you just forgot to let you know. But ideally, will you be here next week with the office hours? Speaker2: [00:00:52] Well, I will here. Speaker1: [00:00:55] So I'm really excited for this. It's going to be such a great event. There's a lot of great topics and conversations being discussed. To me. It almost feels like an science reunion because most of the people on the main stage are either going to be on my show or have already been on my show before. So it's all some kind of family get together from my perspective. But yeah, definitely go and register for that. I'll be sure to include a link to that very shortly. And also in the show notes. Yeah. Super happy to have all you guys here. If you guys have questions, go ahead and put them put. You know, I've got a question into the chat will add you to the Q Well, yeah, ideally, I want to talk a little bit about creativity and creativity as a data scientist. First of all, talk to me about do you think you need to have creativity skills or do you need to be creative as a data scientist? Speaker2: [00:01:44] I think very much so, because a lot of our work isn't really outlined in a straight path, I think, especially because we iterate so much trying to be creative in the factors that we're iterating and knowing enough technical skills to understand which knobs you can change, and then having some creative energy and being able to test and experiment with those and eventually trying to get to a place where you either have a good domain understanding of the types of things that work well and the types of things that don't. But also it allows us to be more than the stereotypical maybe an analyst kind of profile where it's not really seen as a positive thing to have a lot of creativity. But I do think it is really is really necessary for the vast majority of Data science jobs. Speaker1: [00:02:38] Absolutely. I agree with you. I've been really studying and researching about creativity. I go on these interesting, kind of interesting to me, actually, an intellectual type of sprints where I focus on something that I'm just really interested in. Before it was all about luck. I was really fascinated with what is luck. And now it's mostly been shifting towards creativity and what is creativity. How can I be more creative? Because I think it plays an important role in our jobs as Data scientists, because, I mean, yeah, there's analytic work, but at the end of the day, we're solving problems and we need to apply judgment to a problem that is in front of us. And, you know, coming up with possible solutions is very, very creative. And something I've been reading recently is this book right here, a whack on the side of the head by Roger Vannak. I think that's his name. I really, really enjoy this book. So much so that I bought his other books. And one of them is called A Kick in the in the kick in the seat of the Pants, which is interesting that he's got another book that's all about the philosophy of Heraclitus and how Heraclitus can teach us to be more creative. But in this book, he talks about Tom Hirshfield, rules of thumb and two things that are really, really important. Two or three things that are really stood out to me was, one, never state a problem problem to yourself in the same terms as it was brought to you. If you don't understand a problem, then explain it to an audience and listen to yourself. And don't mind approaches that transform one problem into another. That's a new chance. So I found those to be really interesting bits of wisdom there. But I definitely highly, highly recommend this book. And, you know, hopefully I can synthesize everything I'm learning and apply that to to my job as a data scientist. But so talk to me about some practices that you have used when you're up against a problem and you're like, oh, my God, I just I can't figure out a way forward. Speaker2: [00:04:33] Weirdly enough, I have to take a step away. So sometimes this means diving into random creative work that has nothing to do with my day job, where sometimes it's been sitting in maybe with different stakeholder groups. And I'm used to talking to you, but I I've noticed in my kind of workflow I need to either get a more broad view most of the times, and very rarely do I need a more narrow kind of close up view. But in those cases, I try to either look at the data in a different way or have a better under. The standing of the problem itself. So I think that that is kind of where I've fallen back on, like maybe I don't I'm not grasping the problem the way I should be or the way I should attempt to solve this for the stakeholders at hand. So I try to kind of step back a little bit, try and get a wider view of what's going on. And even then, you know, looking outside my own organization, trying to see other solutions to problems in similar industries or anything that could possibly kind of transfer over has been really helpful for me to open my mind. I think the longer you do this, the easier it is to somewhat settle into specific models that you have preferences for. But really, just staying on top of new research, even if I have the chance to read a paper about a way to utilize a model that I haven't tried before, that can often spring up some ideas about how to solve the current problem at hand. I'm kind of pinning a lot Speaker1: [00:06:13] Of the love that some really, really great tips and advice, and I'm definitely will be taking some of that with me. That point about getting up and kind of walking away from the problem, I think is a super important one. Like sometimes I'll just go for a walk or go do some dishes or something or just, you know, stretch out for a few minutes, just let the mind wander and then get back to the problem. And sometimes you have some new insight. But that great. Thank you for sharing some of those tips. Hopefully you guys enjoy that. If you guys have any questions on that, further, let me know. But in the meantime, we've got a question here from Segun. Go for it. I. Hi, how's it going? Yeah, yeah, Speaker3: [00:06:50] Yeah, yeah. How's everybody doing? Joining a little late. Speaker1: [00:06:55] Yeah, yeah, yeah. Man, I'm glad you're here. Thank you. How can Speaker3: [00:06:58] We help. Yeah. A quick question really around the technical part of our interview. So I would like to pose this to everyone that will potentially interview or was interviewed before even interviewing an engineer. And you get a technical part where you have them write code. For example, the hacker could kind of code. And I'm wondering which would you prefer the candidate that writes the code and work on Python shut down or just go the long route? For example, you give in to areas containing four numbers each on the question to combine the two areas together and shot them in ascending order or something. You go the loop route, get the first idea, secondary loop at each using the index and all that, then. Right. Some sort of a certain coded game to store the numbers. I know all you use in both function that Python does have, for example, noncredit. I read the extent of the secondary to just doing automatically the extent that you can use by the array, not sort automatically on the list. So I'm wondering which would you go for? Because I think I would want should walk focus on going to longer route, which involves quite a lot or shorter. I get the work done. I just my Speaker1: [00:08:26] Question, I mean first it sounds like the big goal for that would be super large because there's a lot of steps looping through two arrays and then having to sort them. I mean, I would go for the candidate who is able to write code that is able to execute the fastest and utilize and good principles for doing that. So I don't know if that helps or not. But I mean, I'm not a computer scientist by any means, but I would say probably in the case like that, maybe merge. So it would make the most sense or maybe some type of divide and conquer technique in that sense, ideally. What do you think? Speaker2: [00:08:58] Yeah, I think from my perspective, at least when I've been interviewing, I care a little bit less that maybe they know they can use these loops to kind of get the same answer and more want to see them leverage things that like those inbuilt functions, especially like I mentioned, especially when it comes to timing, complexity, it would be overall slightly still better for the team to bring someone on who is able to write fairly fast code. Considering a lot of what we're doing in Emelle, we're concerned with the optimization and a little bit less with knowing specific techniques to get there and more so building things properly so they go faster. So I'd agree. And probably the second method take the shorter route. I'm sure some, like technical interviewers and H.R. managers, might have a different opinion there. But even that could be something you bring up saying you are in that interview. You can say, you know, I can go this really long route, but I'm going to choose to use in both functions because in the real world, this would be a little easier. That's helpful. Speaker1: [00:10:09] Yeah, thanks. Yeah, because at the end of day, Machine Learning Engineer definitely is a software engineering type of position. So leverage best practices for software engineers and show off the fact that you can write code that is going to be performant. That's yeah, I agree one thing. Did you have an interview like this? How did that Speaker3: [00:10:29] Go? No, I wasn't coming up, so I'm just trying to prepare for it. Speaker1: [00:10:33] Yeah, well, good luck. I'm looking forward to hearing something about that. Right. A shout out to everybody else in the room. We've got Jim and Prete. Dave Thomas Tó. Happy to see you guys here. Anybody have any questions? Go ahead and let me know. You can put it right there in the chat. But right now, I would like to get some feedback on her resume. So let's go ahead and and do that. I wonder if are you able to share your screen? I think I would. You might have to grant her permission. Speaker4: [00:11:03] Yeah. Can I see a green button on the bottom and maybe give it a shot? Speaker1: [00:11:08] Yeah, give it a shot. See what happens. Speaker4: [00:11:10] Ok, so here it goes. Oh, OK. Speaker2: [00:11:13] Looks like the permissions there. So you should be able to share Speaker4: [00:11:20] Something showing up on see. Yeah, no problem. All right. Can you guys see something here. Speaker1: [00:11:29] It's it's coming up. It's your screen. Sharing is starting so excited. Take a look at this. All right. Go ahead and scroll up to the top and zoom. Yeah. And just a little stop. Speaker4: [00:11:40] So I'm either I put you on has Data scientist, but I'm also opening up other areas like Data product. I noticed this is just like the skeleton or whatever you could call it. So I have two projects that I put over here. One is the sentiment analysis and exploratory. One is an R, one is an python. The one on the top sentiment analysis is using Python. One in the bottom is R, and then the rest of it is all my education profile. And then it's got a little bit. At the bottom is other data projects that I have on my GitHub. I can stop here and then I can show you. Is it just one page? I know I do have my work experience which comes out right in the bottom right there like that is my current most recent job. Currently I'm working and the rest of the stuff past past experience. Speaker1: [00:12:37] Yeah. So as much as possible. I mean, try to compress this into one page if you possibly can. So scroll up back up to the top. OK, and if you can you zoom in a little bit more. Speaker4: [00:12:49] Sorry. Say that again. Speaker1: [00:12:49] Can you zoom in a little bit. Speaker4: [00:12:51] Oh yeah. Yeah, yeah. Sure looks I get. Speaker1: [00:12:53] Yeah that's good. Right there. So the profile part let's start there. I'm a data scientist and I would probably move ampersand to just spell out and and uh I'm sorry. Speaker4: [00:13:03] Oh this one here. Right here. Yeah. OK, gotcha. Speaker1: [00:13:06] So let's make that ampersand actually a word. I'm a data scientist and artificial intelligence professional who is passionate about the intersection of eight Data science people and business by delivering key Data different insights drive. So I mean, the profile doesn't really it doesn't make a whole lot of sense to me because about how it's structured. I'm a data scientist and artificial intelligence professional who is passionate about the intersection of A.I. data science people and business by delivering key data driven insights, drive strategies and improve business decision making. So I think just the structure of that sentence, it kind of jumps from Tugwell being passionate and then you're delivering something by doing something like that. It also seems like a disconnect, like to sentence the two parts of a sentence are just kind of stuck together. So maybe find that right. Speaker4: [00:13:56] So maybe just say I am a Data science professional who is delivering key Data something, maybe get rid of this whole thing or. Speaker1: [00:14:05] Yeah, well, I'm not going to go through sentence both, but definitely just think about how you can like what what are you actually trying to tell me with this profile. Let's run that exercise. Like what. When, when you have somebody read this profile, what impression do you want that person to have about you as the candidate? Speaker4: [00:14:20] I want them to hire me as a data scientist that I can I can do the work as a science, you know, like maybe look at that. I'm able to look at the data and provide insights. That's basically what I want to think of me. Speaker1: [00:14:33] Ok, so think about this. The thousand other people who are applying for this exact same job, probably seeing some variation of the exact same thing that you are trying to save the profile. So how can you make yourself stand out in literally twenty five words or less? I think be creative with how you see this. Right. And think of what you can say to separate yourself from the thousands of other people applying for the same job. Right. So I mean, you're probably not passionate about the intersection of ideas and so know if anybody is passionate about that, but they are probably passionate about helping other people solve their problems. Using blah, blah, blah, or maybe passionate about helping improve people's lives through something, something, something. OK, I'll pause there. I'll see if he has any input as well. Speaker2: [00:15:17] Yeah, I think this is also a great spot for you to speak a little bit directly to the hiring manager and say, obviously, you have some Data science experience, but you're still at that point where you want more. You want really to dove in and have maybe a little bit more responsibility. So trying to find creative ways to express that in this profile, saying I'm ready to like take that next step because I've proven I can provide data driven insights. I think that might be a good way of framing that profile section. Speaker1: [00:15:53] Ok, gotcha. Yeah. So something like a I would just say Data science professional writes a Data science professional with a proven track record of something, something something using something. Right. Does that make sense. So that's what I'd ideally saying that about making it a value proposition. OK, gotcha. And so this is a great format for the resume. And this is what we provide that data scientist dream job as a resume template. And as you can see here, it's very nice and eye-catching. So I like that a lot of white space and everything. It's quite good to see. So the sentiment analysis model on new movie reviews. So, you know, huge fan of using the star format. I for one typically don't I don't use bullet points too much just because I find it too constricting. So that's just a personal take on that. I would much rather just use a narrative where I'm saying situation TASC action result. Right. So for this project, I was interested in doing this thing. My task was to do these three things. The actions I took or the analysis I performed was this. And as a result, I observed this. Right. So maybe something like this could be you know, I was interested in determining whether I can accurately classify the sentiment for movie reviews using PI Torch and blah, blah, blah. I performed these following tasks, did this analysis, and as a result, my model performed this compared to baseline the stuff. OK, yeah, ideally I'd love to hear your input here. Yeah. Speaker2: [00:17:34] I actually think, you know, you can either use the Stane format. I have my resume in bullet points, but I'd say you can in some ways attempt to condense this. So I do think you have really good detail. And you've mentioned a lot of the things that stand out, especially in the two of you are going through it like it's done. They'll look for things like AYSO, Sage Maker, LSM, a lot of the key words you use. The only suggestion I'd have is try to like shorten the sentence just slightly if you stick with the bullet point format. I think that is also another reason. It's not necessarily the best way to go because it makes it look so much larger and it looks almost more like a wall of text, even though you have them specifically outlined any a lot of people who review resumes are only able to like read one or two and then they're like, yeah, I'm not going to get the bottom of each one of these. So. Right. Just given the amount of time they spend on your resume, if there's ways you can shorten these somewhat, that would be helpful. Speaker4: [00:18:41] Ok, sure. Thank you. These are all valuable and Harp you said you don't prefer bullet point. You put pretty like in a paragraph format. Speaker1: [00:18:49] That's just me. That's just like everyone's going to have their own thing. Right. Like but there's no there's no like one right way to do a resume. Like you get enough resume advice, everything's going to cancel out to zero. Right. So. Right. What is going to be true between what I ideally myself have said is that, you know, condense it, make it consistent to make it clear that one bullet point follows the next the next day. So, yeah, when you have yours done, I think, like, the flow is great, the flow is great. Maybe one tweak you want to do is have a opening sentence that's free of bullet points, that provides maybe seven to ten words of context. And then you can have the bullet points outline your your individual tasks. OK, so that's something to think about. But I'll be bullet points. The points really doesn't matter. It's personal preference I think. OK, yeah. I just I, I like narrative forms and things like that, but I like your flow is it's nice. Like I said, I do see that flow like you're talking about. You're the second bullet point is the tasks and your actions and the third point and then your results. So you are far following star format. If I was to have this resume on my desk and you are applying for a role like I wouldn't disqualify you, but. Oh, right. It looks like this a really interesting project. Looks like she's done work. So, I mean, that being said, it's. Well done, I would go back to that profile section, that's for sure, Speaker4: [00:20:12] And figure out a way. OK, I will do that. Speaker1: [00:20:16] Yeah. So definitely for more in-depth, you know, your part of DSG, go talk to one of our stronger officers. We'll get more in depth or posted to our channel there. But we'll circle back around if there's if there's time. Right now, we've got a Manpreet in the queue Speaker4: [00:20:31] So I can always come back. Yeah, I could do this another time. That's fine. So I'm going to stop sharing my screen. Yeah. Speaker5: [00:20:38] No, I had the question related to Syria and Speaker4: [00:20:43] And Speaker5: [00:20:44] So I had this goes into being of Russia. You generally don't have the experience you need to show about your project. You did it in your college days and graduation. So at that time, elaborating a project about your project, it's a good idea of keeping it like in a simple way. I did train this and got the scores and they decide to use this and gave it up like Richard elaborated more Speaker1: [00:21:15] I mean, as much detail as possible without taking up too much space. That's my general advice there. Uh, what do you think? Speaker2: [00:21:24] Yeah. Along the same lines, is that providing the detail that is important to people reviewing. And so the things that are important are what kinds of languages, tools you used that's have to get past them and the like applicant tracking systems and the other details that are important, especially outcomes and how you worked on this project. And many of these projects are going to be individual and that's fine. But if you were collaborating with someone else, it's important to mention that as well as providing some context to how the project came about. So for a lot of our personal projects, we just have interest in something. But I think providing some context of I really love Star Trek, so I downloaded all of the scripts from every episode. And even having that, that kind of shows, OK, you chose this project because you were passionate about this or interested in this, I think is mentioned as much of those details you can include like Harpreet Sahota in as little space as possible. Speaker1: [00:22:30] And as a you know, somebody who doesn't have work experience projects, I think are a great way to signal that you have interests in the field, that you have capability to get the job done. So definitely do projects and and talk about the your on your resume. And I would I'm a huge fan of having really interesting titles for projects. Right. So the more Eye-Catching and interesting you can make the title of your project, I think the better chance you have that somebody can read the resume because you have to think of the resume as what it is. It's a sales document typically. Ultimately, it's going to be viewed by human. You know, you can bypass the applicant tracking system by reaching out to people directly on LinkedIn, making sure you're reaching out to recruiters and H.R. people, not just individual contributor data scientists, but then leverage some some of the common biases that humans have. Right. So if I have a resume that's just a wall of text, that's just going to it's going to tire my eyes out. Right. So having whitespace, making it easy to scan and then those buildings that you do there, J that's great because it catches people's attention with personal projects. The more interesting you can make the title and the more interesting you can make that look like talking about why you're doing this project, the better it will turn out in the long run toward do you have any resumé tips here? I mean, I thought Tour's got got awesome advice for you. If you have anything to show you, I'd love to hear it today. Speaker3: [00:23:52] I don't have really any TV. I agree with a lot of the points that you made. I keep it short, make the point, keep it interesting to trigger people's interest. Me personally, I haven't updated my CV in twelve years. And when you get to a point where you stop updating and you know, I'm not really looking for work, but still, you know, I still update it and but it's getting shorter and shorter as I grow older and older. And, you know, at the end of the day, it's kind of becoming self explanatory to a lot of the people I talk to. So there's not so much promotion. But still, I agree the key is eye catching. But the one thing I've started to be more interested in is that when you're looking at applications now, a lot of the reviews, the first reviews is made by A.I. or Apple or reviews automated reviews. And to me, I think that the KB's nowadays really has to be made so that the machines can read it. It's not really for people, it is actually for machines. So getting those trigger words in there, I think. Is key, I'm not too familiar with how those tools work, but from what I see and when I talk to people on LinkedIn, you see a lot of people are playing, but basically you get no response or no reaction. And to me, I think it's a lot to do with the kids now being read by machines and not by people anymore. Speaker1: [00:25:30] Yeah, I mean, one way to bypass that is reaching out to people on on LinkedIn and making sure you're targeting like a technical recruiter who's active on LinkedIn. So you got a little bit of sleuthing. So if you've applied for a company that looks really interesting, go to that company's LinkedIn page, look at the people type and recruiter in terms of like the job title, find people who are recruiters, specifically technical recruiters, click on their LinkedIn profile and and see how active they were. Right. If somebody hasn't posted anything in months, don't waste your email credit messaging them if somebody's been active over the last few days. Definitely. And the credit there and I think that in itself can help bypass a lot of that stuff, because if you just reach out and say, hey, you know, just for the record, I've already applied on your company website. I know that my resume has a unique format. So I, you know, make sure that it didn't get passed on by the system. So I'm just reaching out to you and just be kind and send a nice message. I made a post a while ago about five tips to make a resume stand out. So go ahead. I just read that here for you guys. So five tips to instantly improve the resume. One stand out from the crowd. Hiring managers review thousands of resumes and they all look the same. Simple cognitive hack immediately catch the reviewers AIs. Speaker1: [00:26:47] If every resume looks the same and your stands out in a good way, the reviewer will automatically pause and spend more time looking at over. So we saw that demonstrated which I as a template, really visually appealing. Um, one thing I would say was that which part of the next point, which is about leveraging whitespace rights or wall of text, can overwhelm the reviewer. So cognitive haak use white space to your advantage that I need to rest when you're scanning or reading. So white space is going to help the reviewer digest your resume. Content and separate information out helps create a nice little hierarchy of information as well. Craft a story. So this is where our advice will cancel out to zero because I say don't start firing off bullet points go through you to connect with you. You do that with the story. Paint a picture of your main responsibility. So an opening sentence, maybe describing your role, what you did, and then. Yeah, definitely move into bullet points. Right. Use bullet points to present achievements and highlight projects you've worked on using the star format to be audience centric. Your resume is a sales document. Tell the reviewer what you could do for them. Avoid jargon, be clear and project titles matter. Right. So that's that's kind of the tips that I had there. Um, definitely try to leverage those if possible. But these are tricky, man. Like, they're so they're hard to get. Right. Speaker4: [00:28:09] Um, but yeah. Thank you, Harp. That was very helpful. I work on their profile and, you know, kind of shortening it a little bit. I could I could tell a lot of sextape. I just, you know, get it. Not good enough. But yeah, I think I know. Speaker1: [00:28:26] What other advice would you share regarding resumes? Speaker2: [00:28:29] Um, I would say I guess it's a little bit less about resumes and more about projects. Try and gear your projects if you are interested in certain industries. And I know some people who are, because of their past, super interested in working for airlines or super interested in transportation or finance, try and work on projects in those and then it is going to be easier. And you'll probably find if you make it on to the later stages in the interview process, they'll want to talk to you really in-depth about anything you've done that overlaps a little with the kinds of work they do. So based off of your interest, try and find target industries and then try and have a couple projects based around that. And I would double down on having creative project names. As someone who has reviewed a couple hundred Data science resumes, it's easy to see some of the same projects reappear more or less in this in the similar words. So even like the I would say maybe is there a Data a movie in that Data said that you love or you really hate and then call that project? You know, everyone else hates this movie like idee or something. And so those are ways that kind of pick the reviewers interest as well. And so they can also ask, OK, why do you name it that in addition to what makes you care about the industry, you're working for us specifically. And if you have a project that's similar in the same industry, that becomes a lot easier to answer Speaker1: [00:30:05] As an awesome tips. I like that. That's when you're going out and you're trying to think about a project that you want to do. How do you go about editing on a project? Is it kind of doing something that you're just really interested in or could it be, you know, I've got this ideal company I want to work for, and I know that this ideal company works on this type of problem. Let me try to emulate that and then just show them that I did this. I mean, because there's so many different ways you can do this, right? Like think about it. You're only limited by creativity. If there's a company out there that, you know, maybe is in your local municipality, in your local area, that you're like, dude, I would love to work there, but they don't have any job openings available, then one way to get your foot in the door is just create a project that is kind of touching on the type of work they do, email A or LinkedIn message. A data scientist obviously build a relationship with them and say, hey, by the way, you guys are doing cool stuff. I find it super fascinating. I was just trying to replicate what you guys do on a daily basis or whatever. Obviously, I don't know what you guys do, but here's a project that I think you might find interesting. Check it out. And you just kind of you kind of just introducing yourself. And then when an opportunity comes up, then maybe they'll reach out to you or maybe you can create an opportunity for yourself or one didn't exist before. That's something I was doing for myself quite a lot. Back in twenty eighteen, I was just message people and create opportunities. How have you created some opportunities for yourself. I really Speaker2: [00:31:27] Yeah. I can say that I've definitely taken that path trying to build relationships on LinkedIn. It's not necessarily easy, but especially if so, one of the ways I found to make it easier to do a lot of digging and try and find anything outside of work that you have in common with people. So I have a ton of random interests, like I'm really into hockey. So anyone I've connected with that is remotely into hockey. I'm like, OK, I have something to talk to you about other than like work. So it might come from finding their personal website, finding their Twitter account, seeing other things that they're interested in to build a stronger relationship. Then as you can imagine, the larger the company they're at and the more name recognition. So if there's someone who out there doing talks, they're consistently getting these kinds of messages of, hey, you know, I'd love to work for your company and it's really hard to stand out there. So by trying to find even if it's something I am, like, not super interested in, but I can have a conversation about, like I currently live in Denver. So there's a lot of people who are Data scientists here who are really into skiing, snowboarding, the outdoors. I can say, oh, you know, I heard about the powder and it's on like the mountain like and I'm not a skier snowboarder, but I can kind of start that conversation that way. So I've tried to start in not necessarily work related conversations, but trying to find people who I think would be those good connections and trying to find anything we have in common that's not just work. So that's been helpful for me. Speaker1: [00:33:07] That's so cool. I didn't know you're into hockey is like a Canadian there. So Dave has a question here. It's a good idea to tailor your resume to the company that you're currently applying for. That's a great question. Um, so so me personally, like I'm lazy. So I would I just I would I would spend my time more trying to find people to reach out to at the company that I'm interested in, then optimizing for just the resume itself. Because ultimately the purpose of the resume is just to, you know, a sales document, try to get try to get yourself noticed. But you can also get yourself noticed by just reaching out to people, recruiters in the company and talking about how what you've done connects to what this company is currently doing. Right. What do you think? Speaker2: [00:33:54] Ideally, yeah, I think you can. But I also agree with you. I mean, I'm lazy and I'm not doing that for every company, but I do have specific, like profile recipes. I have my resume if I'm going out for, like a deverall or if I'm going up for a Data Science Lead role or if I'm going up for a like product analyst kind of role. So I have then kind of categorized in by by that what kind of job profile, regardless of actual title. So you'll see some jobs that are business analysts, but they ask for a data scientist essentially, and the requirements and I submit my data scientist like a resume versus the one I have maybe prepared for more analyst roles. So that's what you can do it and save a little bit of time. So it's not especially applying to jobs is really difficult. And I know there's so many times that you have your hand in a couple dozen applications at once. So if you can at least have maybe two or three that are your goatees, that might be an easier way to manage it. But still tailor these. Resumes and you can also highlight different projects, depending on what kind of job you're going up for. So I think that that's been a been a good way for me to try to manage that without having to individualize a little too much. Speaker1: [00:35:21] I like that. I like talking here. It's like having multiple profiles like that would be like that, that approach. That's one thing I found interesting there was that when you talk about product analyst and I'm curious like what happened to Lillian Pierson about this when I was interviewing her for the podcast couple weeks ago, maybe this last week or something. But we're talking about how to move up in your career as a data scientist, because there's rules where you just a number of means they just stay. But you're an individual contributor type of role. So very much hands on, you know, in the dirt, in your notebooks, in your scripts and just doing doing that work. But then if you want to get onto more of a leadership and strategic path, then maybe at some point want to change directions. So how do you see moving to like a product analyst or product management position, setting up a data scientist to go on to leadership positions? Speaker2: [00:36:15] At least from what I've heard and because I haven't done this myself, I think it's actually a really great jump in that the vast majority as like program or project kind of management roles, while they don't require you to be on the Harp like in the dirt, in the weeds and technical, they require a lot more similar skills to leadership when it comes to use strategy and when it comes to time management organization of projects on the entire scale. So in most Data science roles, we might be working on the end to end for a specific project. But we're not really considering how it impacts the overall organization and the amount of paid men hours that go into it. I think that I've seen people that successfully move into like Data science management and reading. Really large teams have spent some time doing really deep work on strategy as well as just kind of leadership. So some people maybe take the Data science team lead kind of position before they move on to a more people management. But a lot of the success, I think, comes from looking outside of the detail and being able to work with the big picture and being able to communicate the big picture to various levels of like technical skills to your people that are in your August. Well, as other members of leadership. And I think that's what a lot more program managers technical problem injuries tend to spend their time on is communicating timelines, priorities to CEO level, to, you know, machine learning, engineer level kind of people. And I've noticed that they end up being really successful as managers and as leaders by being able to be more strategic. Speaker1: [00:38:14] That's the clouds part, the clouds up there, big picture. And then we have this powerful to be able to be simultaneously in the in the clouds and the dirt, be able to understand a little bit of both. I was actually I interviewed just for the fun of it for a technical product manager role recently got rejected for it just because I clearly wasn't a good fit for it. But I wanted to see what the interview was about. Boss, if you're listening, you know I love you and I love working with you. So just add that caveat that I'm not actually looking to leave. I just I like entertaining. It's fun, but I you know, we had a great conversation, this great conversation. But I knew it was going to be a good fit for the role. So that's what Tom said. I was like, look, I want to at least learn something while I have you here. So talk to me about us product manager. We're talking about, you know, prioritization and things like that. What do you use how do you go about setting these priorities when you're working on a new product or a new feature, what have you? And he told me about this framework, the right framework. Speaker1: [00:39:14] Ah, I see. I'm not sure if you guys have heard of that, but it's theorize prioritization framework for product managers and, you know, it helps estimate value and things like that. But I'll go ahead and I'll put this into the into the chart here. So you guys are interested in checking it out. You can do so. Have you heard of this at all? No. Yeah, it's been interesting. I haven't I haven't read too much into it. That's something I've been trying to I'm trying to give myself an MBA recently by just studying different topics. So one thing I've been like Lean Analytics is is a great book that I've been using, kind of understand products and how metrics we should be tracking metrics for certain products. Written by Alison Kraul, who is also on my podcast Once Upon a Time. Check out that interview and this book right here. Case in point, just trying to understand how consultants think through. Projects when they're just starting out, so they've been helpful in trying to understand more business type of stuff. What do you think? Ideally, have you been have you have an MBA? Right. Speaker2: [00:40:18] If I leave it or not, I don't sell my my business experience. And pretty much all of my business acumen comes from working at startups. And it is a great, difficult way to basically get a like mini MBA. So I guess I should start from the beginning. I when I started working at startups, I actually went through a startup accelerator and that was like a mini MBA because we were forced to pitch our products and go through all of the steps of trying to set things up. Finding a CFO that taught me, I think, the vast majority of what I needed to understand how Data fits in, because as I was going through this business accelerator, I was in grad school for my Data science degree. So I was able to clearly seek some the connections and get exposed to all of the aspects of business I like had no idea about. So I had previously been working in marketing and yeah, you kind of hear a little bit about sales functions and what they're doing, but I wasn't really involved until going through this accelerator and then trying to leverage everything I was learning as like a baby Data scientist at a startup. So yeah, not a not formally trained really, really in business, but having a quick dove in and then having immediate responsibility, I think definitely put some pressure on me to make connections between me, Data kinds of things I was learning. So understanding how to use different kind of regression modeling to work on churn and other product analytics. So yeah, no MBA here actually, Speaker1: [00:42:08] But that's we've learned so much of that through those type of experiences are so cool because like the the I mean clearly very knowledgeable connecting their sides to business. And it's always awesome to hear you talk about that. From your experience working in startups, what do you think would be like the biggest lesson that you've learned that you think all of us sitting in this room should should take with us to? Speaker2: [00:42:31] That's a good question. While it sounds vague, I would say things are not always as they appear. So especially I actually read really good advice the other day where I was like Data. Scientists are not like servers at a restaurant. You don't aren't you exist just to take orders and then go find things in the Data to match these orders. You should be more of a trusted advisor. So you are allowed to have your own morals to specific lines that you're not going to cross within the kinds of modeling you're being asked to do. And it's OK to surface those. I think being able to communicate well and being able to communicate tactfully why maybe you should go in a different direction is really, really important. I think. So just being being incredibly critical and skeptical sounds in most scenarios really negative. But I think one of the strongest things that really great data scientists have is the ability to say no and have an ABI have a bias toward saying no. It's more likely that the exceptions that get passed, you were actually really good use cases and try and see yourself as somewhat of the filter between every, like, harebrained idea someone wants to do in your startup and things that are actually going to provide value and provide value in the right ways. Speaker2: [00:44:03] So with optimizing anything we can over focus on metrics and we can try to optimize for metrics so much that we end up doing the opposite of what we want to do. So there's like a way to measure how effective teachers are by looking at student test scores, but that ends up promoting cheating and a lot of other things so that teachers don't have to worry about job security in case that students don't test well. So just an example of what that overall optimization looks like. So just in general, be skeptical and er on this on the side of their telling you that they want something, but that may not be what they really want and do spend the time to do the digging to try and provide solutions that answer what they need and not necessarily what they asked for. That makes sense. Speaker1: [00:45:00] Yeah, it's almost like a kind of exercising good, good judgment. And in a sense it's not like. That just cultivating good judgment and acting on it. I like that some really good advice. Thank you. So for those of us who really like, for example, let's say we don't work at a startup and you probably don't see any opportunities for us to to jump in and make our own startup, how can we develop some of these Common Core business skills? I know there's not like one skill that's called business. There's really a skill, just like the assizes, a motorcycle. But are there any books that you recommend or anything you recommend to to help us kind of develop more of a business acumen? Speaker2: [00:45:40] Yeah, I think I may have recommended this before, but I really like Data science for business in that it puts a lot of these in really a lot of the problems that businesses are actually facing in terms that Data science scientists can understand. But also it shows you how to start seeing regular business issues as Data problems and as potentially messy machine learning solutions. Other resources. I would say I don't have any specific on deck. However, I would suggest that whatever industry you're working in, try and find and read any of the recent papers that are applying machine learning to any of the kinds of Data in that industry. While the papers might not really give you the business acumen, it can kind of help show you the the path to a technical solution. And hopefully you have enough understanding of kind of the domain already. You can start forming those connections. I also suggest just kind of looking for even like really good medium post about product analytics. So I know that the vast majority of data science is not just in product analytics. I think it's just one of the most popular areas and areas that most fēng companies are incredibly interested in. So understanding what the product metrics, metrics are, say what is churn? What is a you know, what are the metrics that SaaS companies are interested in? Having an understanding there makes a little easier to find. There's a lot of resources there on track attempting to use statistical modeling and machine learning on those types of problems. Speaker1: [00:47:36] Yeah, yeah, definitely. Like products are, you know, it makes sense that that's huge for a producer. And I think as we begin to move, you know, obviously we're always moving into the future, but these type of products are what we're going to just continue to see more and more and more of these type of products that are digital like software based products. So having understanding of what drives revenue for them and how to minimize costs, I think Superimportant said there's Data center business. Great book. Lean Analytics is something I like as well that recently been finding myself going back to readjust to to stay smart and educated in that area. But how about like this? I'm just thinking like entrepreneurs, perhaps like a founder kind of mindset. So in your experience with startups, did you feel that you had to kind of adopt that founder mindset? Speaker2: [00:48:23] Absolutely. I had to have some of the skin in the game to have a really, I think, a meaningful impact. So when I see that, I mean that I was incredibly invested not only because this is my company and I have equity in this company, but I was invested because when you are in an organization that small, you can easily see the decisions that lead to sales revenue and some of the decisions that don't lead to that or that lead to some of that stagnation. So if there's a massive feeling of the film when you can create a model at a really small org, and it's incredibly impactful and they're using it for years and years. So I had to I was a little bit forced to to adopt the founder mindset by being technically a co-founder in the beginning. But it even other startups I joined where maybe I wasn't in the kind of founder level, I had to understand that certain decisions, certain models are coming from places of trying to advance and grow, especially for growth stage startups. So I wasn't able to it brings its own set of challenges. I wasn't able to. You have the kinds of budget necessary to get all of the data I need. Speaker2: [00:49:51] So I had to figure out ways to work with what I had, figure out ways to be flexible and maybe secure a little bit of budget for a little bit of additional Data instead of what I originally wanted. It's also made me more flexible in finding some. Asians under tougher constraints. So you have the constraint of money and time and resources, like my very first like Data scientist position, I was having to extract metrics from screenshots like it seems ridiculous. And this was before I even knew about a lot of OCR tools. So having to do some of those things manually while it was a slog, it did also force me to be more flexible in how I could solve a problem and spent more time solving little problems and then sort of readjusting our overall, like Data schema so that we wouldn't run into the same problems over and over again. And that wouldn't be the Blocher to a new project that is supposedly going to be highly profitable. So a lot of, believe it or not, Data engineering work is involved, especially at that kind of startup level experience. Speaker1: [00:51:09] Let's talk about something that will drive creativity, its constraints that will force you to think creatively. I'm curious. I don't know how much you're able to discuss this. Let me know if if you're not. But you were talking about having secure budget to get Data. What did you mean by that? Um, I get Data from, like, your own sources or get Data from external. Speaker2: [00:51:31] Yeah. So in my scenario is getting Data from external sources. So the start up I was at was about six people so incredibly small and we did not have a lot of data outside of our own transactions. So what we were looking for, there were also organizations that sold Data in this space, but it was really, really expensive. So if I wanted to go and get like two terabytes of industry like transaction data, they were quoting us close to like fifty thousand dollars. And the startup is like, well, that was our last round. So we're going to have to figure out other solutions. Essentially brokered a deal to literally get like half of terabyte of data before, I want to say was about a five thousand dollars. And we ended up doing some analysis work for them. So we actually, as two organizations did a little bit of business trading and trading resources and services. And that's how we ended. I ended up getting enough data to start making initial models and to start building data pipelines for when you were able to get more. But it took a lot of work that I don't think I expected that I expected, as o you know, hey, I'm a data scientist. I can just go ask for this data and I'll be given the data. And that was not the case. Speaker1: [00:52:55] So this data that like, that's interesting. So they had transactional data like where do they get this data from? What was it they just make it up or how? Speaker2: [00:53:05] Yeah. So this was actually and want to say around twenty sixteen. So this was when Colorado actually fully went recreationally legal on like cannabis sales. And so that data was incredibly private and really just held by the state government and the point of sale systems. And so the organization that actually ran in Sobeys is going to sell systems turned into a data and analytics company because they realized what people were more interested in was how much is the average sale, how much should I what should I charge for X amount of product based on all of the other companies already selling it? And so our company was a was attempting to be a bridge between new products here. So we don't really have ways to advertise because we're in a regulated industry. We can't buy Facebook ads, we can't buy Twitter, YouTube, we can't really advertise in public. So we connected those organizations to private media outlets that weren't really they would either be private like news websites or individual artists who would do this work for them. But we had no clue what the state of the industry looks like. We weren't sure how many sales overall in the state are there? How many dispensaries are there? We we had no clue that kind of Data. So that's what we ended up purchasing from them. To have a better idea of what the landscape looked like overall. Speaker1: [00:54:38] That's really fascinating, super interesting to think about all these different types of industries out there that provide all this type of information like, wow. Yeah, that's just it is interesting to me how there's like just so many industries and businesses and so many ways that you can create a business to do something interesting that you probably wouldn't have thought of if all you were exposed to were just, you know, companies in your local area. Yeah, that's really, really insightful. Thank you very much. I guess we should open up everybody else, ask questions I've been so interested in. Learn about Odelia in what she's been doing, last minute questions from anybody here, Speaker4: [00:55:12] Identify a company and a look at your website or things that people go through to look out for their local company in the area. Speaker1: [00:55:23] Yeah, when I was doing it, so I was I just typed in software companies like in my city, Winnipeg, Google, and a bunch of people came up. Another thing I did was, you know, you can type in like your postal code, your zip code, and you can look for people who have the job title, software developer or software engineer. And we're using that phrase particularly because we're looking, you know, in my case, targeting software companies, because I imagine if this company is in software, then they might have a need for data scientists or Data analysts. Right. So kind of thinking about it in that way. So look for look for software engineers, software developers in your local area, see where they work, go look at their company website and see what the company is all about. And if they have data scientists, great. That means that's a potential for you to to networking and get to know them. If they don't have data scientist, then you send out a message like the CEO or whatever, if it's a small company and say, hey, you know, I know you're not hiring for data scientists, but let me tell you all of my skills. Here's what I could do for you. Speaker4: [00:56:25] So you just do the usual open and open a close call in the postal code and then the software looks like that. Yeah. Speaker1: [00:56:34] Yeah, that's awesome. Um, right. Well, I guess we can go ahead and call it today. Thanks, guys, for hanging out. I'm wondering maybe, you know, at some point the future after we have this round of office hours completed, maybe we might want to look at a different time or day for for this. I don't know. I would love to get more people here. I know if you're listening on the show, I know, you know, we end up usually getting a few hundred downloads of of this within the first couple of days. Guys, feel free to join us. Come, come hang out. Send me an email. If there are other times I work for you, I'd love to hear about that, too. So, you know, I was hoping hoping that this this time of day, this time of week would get a lot of people coming in because, you know, we've got people in Europe, people in India that could join in as well, U.K. and so on and so forth. So if you listen at home, come join us one of these days. Man, what have you guys here right on? Well, ideally, any closing words here? Speaker2: [00:57:27] I'd say stay motivated, don't give up and try in like Harpreet Sahota PhD. If you're looking for folks in your local area, look for companies that have Data engineers, it'll be a little easier for you if you do. And a Data science job there. Speaker1: [00:57:42] Yes. Great, great. Hip, cool. Well, guys, remember, catch us next week at the Data Science School Virtual Conference. Super excited to see you guys there. There'll be a lot of great talks. I'm excited for the panel discussion. There's me, Ayatullah, my good friend Vinicius there, who I know you guys all know, and another individual who I don't quite know yet, but we'll get to know her. So, guys, thank you again. Thanks for hanging out. Remember, you've got one life on this planet. Why not try to do some big. Cheers, everybody. Speaker4: [00:58:11] Thanks. Bye bye. Bye, everyone.