In this episode of the What’s AI podcast, I had the privilege of sitting down with Avery Smith, an expert in data analytics known for empowering the next generation of data professionals through his Data Analytics Accelerator program Data Career Jumpstart. Avery has gathered over 100,000 followers on LinkedIn by sharing his data-driven insights and his unique formula for landing a first data job in just 90 days.
Avery emphasizes the importance of practical experience, advocating for hands-on projects over traditional education methods. According to Avery, the key to standing out in the job market is to showcase your ability to solve real-world problems with data. He gives many practical insights and advice to build this knowledge and experience in this discussion.
Avery’s insights on leveraging AI, specifically ChatGPT, in the learning process are also super valuable. He highlighted how aspiring data analysts could use AI as a learning tool to simulate data analytics scenarios, improve coding skills, and even prepare for job interviews. Avery’s approach to leveraging AI makes it an accessible resource for personal and professional growth.
Moreover, Avery shared his thoughts on the future of data analytics, stressing the importance of adaptability and continuous learning in a rapidly evolving field due to technological advancements. His advice for those looking to enter the data world is to never stop learning and build a strong network, which he also details in this episode.
This episode is for anyone interested in data analytics, AI, and jumping into this amazing field from another background (or for students and juniors in AI). Avery Smith provides a roadmap for achieving success in the data field and inspires us to leverage technology to learn far more efficiently. Join us in this episode as we jump into Avery’s secret formula for landing your first data job in 90 days and how to harness AI for learning and growth in the data analytics realm!
The full transcript:
Avery Smith: [00:00:00] I see data scientists, data engineer, data analyst roles continuing to rise over the next decade. I think it is tricky because a lot of people see layoffs at like big tech, like fang companies and stuff like that. But like fang companies, I feel like they way overhired during COVID. And so I think it's just kind of a reaction to that.
So I try to tell people to not look at those necessarily as, as the shining examples of, of where the industry is at, because there's still lots of data jobs. And companies that aren't tech at all, when it comes to you actually deploying models and having them be used by people, trust is huge in industry.
These humans have to be able to trust your algorithm, but before you, you garner that trust. Explainability is a big deal. If you can explain what's going on, or if you can have explainability in your model that helps the human kind of, Oh, this is what I, this is why it's suggesting this. That can be really big because if you have a model, by the end of the day, the humans making the decision and they're ignoring your model, your model is not very useful.[00:01:00]
Louis-Francois Bouchard: Welcome to the What's AI Podcast. This is your host, Louis FranÃ§ois Bouchard, and today I received Avery Smith, founder of the Data Carrier Jumpstart platform, where he teaches you how to land a job in data in 90 days coming from another field. In this episode, I focused on education, how AI is affecting education, and also applied tips to leverage ChatGPT and other AI tools to learn more, but also practice your skills to be more productive and find a job.
It was an amazing discussion and Avery gave a lot of great insights related to data, productivity, and always learning more. I hope you enjoy this episode. And if you do, Please don't forget to leave a like below or a five star review depending on where you are listening this episode on. Let's dive right into it.
Avery Smith: So I studied chemical engineering in school actually. And after my first semester, I was like, crap, I don't like chemical engineering at all, but I didn't really know what else to, what to do instead. and I got super lucky [00:02:00] and I ended up working at a small. startup that made these sensors that basically would smell the air and tell you what was in the air.
If there was something dangerous, like a drug or a bomb or something like that. And there was a data scientist who worked there and he basically kind of took me over his wing and taught me how to become a data scientist on that job. And he eventually quit and I took over for him. And then I went to work as a data scientist at Exxon and I actually got a master's in data science while I was there as well.
And basically. When I was breaking into the fields, it felt like there was not a ton of really good online resources. To learn, there was a lot of bootcamps, but they were really expensive. and I was like, man, I feel like the best way to break into data science and to data in general is by doing projects and by making it fun.
And there really wasn't very many platforms that were trying to do that. And I said, well, I'm going to start my own and really, you know, create a platform that's really focused on breaking in like the career aspects and the project aspects, because I thought that's what was most [00:03:00] useful for me.
Louis-Francois Bouchard: I couldn't agree more on the project aspect and that's also what I'm trying to build it with towards the eye, but I wonder for data specifically, you have a pretty bold claim of, you can bring the students to land a job in 90 days.
So I wonder first, how do you do that? Like, how do you allow people to land a job in 90 days, but also why? Where are you the right person to build the platform? Do you think it's like some kind of interview question, I guess, but why you for, for building this, this data learning platform?
Avery Smith: I'll tackle the first question first.
And yeah, my LinkedIn says that help people land data jobs in 90 days. And the reason I say that is because that's literally what I do and it's possible. Now there's a bunch of caveats and asterisks on that statement, that you probably don't see on LinkedIn. for example, When I first started this platform, I was trying to help people [00:04:00] land data science jobs.
And especially from like non technical or non STEM or non traditional backgrounds. And what I realized then is that it's really hard to land a data science job, especially if you don't have a technical background. Cause not only are you like switching careers into something new, but you're learning a lot of new math.
And you're probably learning to program. Like it's really hard to do data science without being able to program. And so when you're learning new math, when you're transferring careers and you're learning to program those three things add up and it takes a lot, a lot longer than 90 days, the majority of the time it was really hard.
And so what I've pivoted into more is like BI, business intelligence and data analytics and financial analytics and business analytics. And those jobs don't require as much coding and then it requires as much math. So you can do it. a lot faster and the really interesting thing about when it comes to landing a job is it's actually not that important of how much, you know, it's how do you market yourself?
And the way that you, you market yourself, the [00:05:00] way that you actually conduct your job hunt, that ends up being actually much more important than your actual technical skills at the end of the day. So that's one of the reasons why we're able to do it faster because we focus more on that versus your technical skills.
Of course, we still need to focus on technical skills, but it's not big as big of a factor. And then your second question about like, why was I the right person to build the platform that I did? One, I don't know if I am, I'm still figuring that, figuring that out every day. but two is, I think, When I was in my master's program and a lot of the resources I was consuming online, data science and data in general was boring.
It was like super boring. It was taught by, you know, someone who was older. It was taught by someone who talked like this kind of monotone and explained, or like. Or like for me being in, in the U.S. It was taught by someone who, who, you know, maybe wasn't from the U.S. Or something like that. And I was like, this isn't as fun and engaging as it could be and so that's one of the things I [00:06:00] pride myself on is like making data more fun. I talk kind of fast. I talk kind of animated. And when I'm, when I'm teaching, I'm always trying to do it in a fun scenario. So actually the first time I taught like a data science course or anything like that, it was actually basically right after the pandemic started.
And like, we did like this whole use case study with, trying to predict if there's going to be toilet paper at your local grocery store. And that was just, this is a fun example, right? So that's one thing I think I bring to the tables. I try to make data fun.
Louis-Francois Bouchard: Yeah. That's essential. I live the same thing in university is just, well, you do have some very interesting props.
And I guess that's. This is where you end up going to, like, if you have a nice math teacher, you will enjoy math and you will go towards the math route, like, I, I guess it all depends on, on the professor. And that's also the good side of internet. Like you can basically choose your professor and your favorite learning method.
So that's. Pretty cool. And before getting into the, this more [00:07:00] specifically, you mentioned that if you are going into AI, the AI route, you, you need to learn programming, but you also need to learn math. And maybe that's less true for data analytics, but you will still be using algorithms and other math related things.
So I wonder what's your opinion on learning the theory, like how those algorithms were built or like. This, the math behind the algorithms, is it relevant? Is it worth learning or do you, should you just learn like the, the one line of code that implements it?
Avery Smith: This is kind of a heated question. And I think there's people in both parties, people who are really pro application and people who are really pro theory, and maybe this is another reason why.
Maybe people would like to learn from me or they'd hate to learn from me. I hate theory. I personally have never enjoyed it. I've met, like, even in school, when, when I was learning chemical engineering, they teach you all like these, like the [00:08:00] theory behind it. And, you know, this is the formulas and stuff like this.
And I'm like, great. Skip all of that. Wake me up when we actually get to the application of why this is important. And so that's, that's, I'm super biased to like actually implementing data things. Now, I don't think you should totally ignore theory necessarily. Like it's definitely can be helpful and help you implement better but when it comes to like, if I had to choose one or the other, I personally am always going to choose application. over theory in both my, my preference choice personally, and then my teaching style as well.
Louis-Francois Bouchard: And instead of theory, do you think it's important to also try to explain what you're doing, like making it more accessible?
Like for example, the theory would be the person building the thing, understanding the underlying. But do you think it's a pointer to try to explain if you are using a CNN or whatever algorithm, try to explain it with [00:09:00] your company or with the people using it, how it works in like layman's terms? Like, is it important or should they also?
Not know, like it's not relevant, just like the theory is not really relevant.
Avery Smith: When it comes to, you know, people actually, you actually deploying models and having them be used by people, trust is huge in industry, like they have to, these humans have to be able to trust your algorithm. And so you either, you know, you either can earn that by proving that you're right over and over and over again, you know, and eventually hopefully the humans are like, okay, like I get it. Like this algorithm, this really works. I, I, I've seen the results, but before you, you garner that trust or maybe your model isn't good enough. explainability is a big deal.
so for example, in most companies around the world, although there's algorithms that are like suggesting what to do, a lot of the times a human is still making the decision at the end of the day. And so with that human, you know, if, if the commuter says, you [00:10:00] know, we should go up, but the human's still deciding, are we going to go up?
Are we going to go down? Trust in your model can really make the difference about what that human actually decides. So if you can have, if you can explain what's going on, or if you can have explainability in your model that helps the human kind of, Oh, this is what I, this is why it's suggesting this.
This is probably why it's a good idea. That can be really big because if you have a model, by the end of the day, the human's making the decision and they're ignoring your model, your model is not very useful.
Louis-Francois Bouchard: Yeah. And about education in general, what's your thought on traditional education and online education?
Will it, are they just complementary one to each other or like will online education replace completely traditional education?
Avery Smith: Online education has opened up so many opportunities for so many people. So for example, I actually got my master's degree from Georgia Tech. I've never even been to the state.
Of Georgia. So, that like obviously opened up, like, it's a really world class, technology college in the United States. And it gave me the opportunity [00:11:00] to go there fairly affordably still like $17,000. And I think there's more affordable things that you can do. but still like, it wasn't like $40,000 or $70,000, right.
Yeah. At the same time, there is something to be said about being in person. I feel like it's a lot more engaging to be in person versus online. So I think it's, it's a trade off. It's like, how engaged do you want to be? And like, how much community presence do you want to feel compared to the flexibility of it all, right? Because like, if I went in person, I couldn't, I couldn't do it on my own schedule, which is difficult. So I think those there's, there's still room for both places. If you, if you really want to be in person, you really want to be engaged. You feel like that's important for your learning. I say go there.
But if you're like, Hey, I only have, you know, six hours a week at like random times, maybe, you know, 10 PM to midnight on Tuesday and Friday. Like traditional education might not work for you. So online education might be the right fit. So I think there's room for both of them moving forward and it'll be interesting to see how they interact with one another.[00:12:00]
Louis-Francois Bouchard: Yeah. That surprised me that you are, you did your master's remotely just because I've seen many of your posts talk about the net, like building your network and the. Usefulness of a good network. So I wonder how did you manage to grow and develop a network if you did your master's remotely, like, isn't it better to do it in person and try to make friends and just contacts for the future?
Or you did still manage to make great contacts and along the way, even remotely.
Avery Smith: You know, that's, that's a really interesting question. I was actually reading this book right before I hopped on this call. It's, it's called the million dollar weekend by Noah Kagan. he started, AppSumo and it was actually really interesting.
So this guy's like a multi millionaire and this quote was really interesting. He said 90 percent of my net worth. Comes from meeting people. I thought that was so fascinating for him to say that. and I don't [00:13:00] think that is a hundred percent true for me personally, but I'm also not like, I don't have near net worth that he has, so maybe I need to start meeting more people networking for me has come really from, from.
In person like connections for me. I go to church every Sunday and my church has like a super big network with, with lots of interesting people. And so I'm able to meet a lot of really cool, interesting people through my church. So that's like one way I network, but then predominantly a lot of my networking has come from posting on LinkedIn almost exactly four years ago, I started posting on LinkedIn. during the pandemic at first it was to my first post, like ever on LinkedIn, basically was the U.S. Government put out like a petition for all data scientists to use their NLP skills to try to figure out what we know about COVID, how to fight COVID, how dangerous is COVID.
This is like right early days of the pandemic. And I posted about that. And I was like working on it. I was like live [00:14:00] stream working on it. And I'm so bad at NLP. So I didn't get anything really done. but I tagged a bunch of people on LinkedIn and that post ended up getting 80,000 views on LinkedIn.
And I was like, people are on LinkedIn and they, they see stuff. What the heck? And basically since then I started posting almost every single day on LinkedIn for four years. And I've grown my network to like 107, 000 followers now, and you know, I haven't met any of those people really, you know, but just by providing them value, hopefully via posts and maybe direct messages and comments has really expanded my network quite large to where if I, if I need something, I can often ask my network and hopefully get some help.
Louis-Francois Bouchard: How do you manage to post every day? How do you find ideas or how do you manage to bring value every day?
Avery Smith: It's definitely hard. And I definitely think I'll, I'll say I fail. Sometimes I'm bringing value every day. I think you can educate people. You can entertain people. You can support people. There's different methods, that, that you can do based off [00:15:00] of what your interests are and what you're trying, trying to do.
It's, it's hard. It's really hard. I, I would say another quote here from, I think Pablo Picasso, he says, "All art is theft". And that's one of the things that I do is I try to find stuff that really is. Motivates me or inspires me or teaches me and twist it, give it my own twist, and then, you know, put it in my domain and give it to my audience.
So I'm really inspired by, by other people's. And you know, I, I watch a lot of YouTube and like sing some of the stuff that people do on YouTube. I'm like, man, let's just do this, but make it for data analytics or something like that. So I, I, I'll say I get inspired by a, a lot of people. the other one is, I do, I do kind of use a little bit of AI, you know, giving some ideas to get me, get me started on my posts. It's very, in fact, it's never happened where AI has just created a full post for me that I've actually used, but at least it gives me a warm start.
Louis-Francois Bouchard: Yeah. I do the same for, well, not for ideas, but for like editing or suggesting titles or things like this.
It's, it's, it's [00:16:00] very good. It's a, yeah, I think it's a very good editor if you go back and forth with it. I guess brainstorming as well is great with it. It's, it's really funny because I don't know if you. You follow this, but here where I live, I know that universities really don't like ChatGPT. Like they don't like.
Students using AI, whereas it's like so good and so useful to learn anything. My next question was about the universities and more specifically graduate studies. But do you think they are still relevant now? Graduate studies doing a master's as you did, or even a PhD? Like, is it something that is still relevant now with ChatGPT where you can learn and almost do anything, but all the other resources online?
Is it still worth to go for a two year, four year or whatever degree?
Avery Smith: I think it really depends on what you're trying to accomplish. If you're trying to like, for example, break into data analytics, I don't think it's needed at all. And I think there's better, faster methods to do so that are, that are more affordable.
But like, if you want to network, If you want like [00:17:00] a top tier job at like a really top tier company, if you want to like master the theory behind it, then, then go for it. So I think I almost feel like you're going, it's like, it's not like what path is better. It's like, what destination do you want to end that?
So I still think there's, there's places for it, by the way, you won't believe. So, I had a. I do one on one coaching calls with people who want to break into data analytics. And I always ask, like, have you taken any of the previous data courses? And this person just recently told me that they're in a master's of data analytics with ChatGPT.
I don't know, like, I haven't talked to this person yet. So I've never heard of that. I'm highly suspicious that it's. That, that even exists, but I mean, maybe the master's programs are even, even starting to change too. Who knows?
Louis-Francois Bouchard: Wow. And for data analytics specifically, then what would be the, like you mentioned, there are other ways and better ways than university.
So what, what would be the skills that you need for [00:18:00] data analytics? What, what do you need to learn to become like a professional or freelancer or just work in the field?
Avery Smith: Once again, I really like the idea of starting with the end in mind. So if you want to be a data engineer versus a data scientist versus a data analyst, so on and so forth, you're going to have to have different skills.
So I really help people become like business analysts, financial analysts, data analysts, and that's another thing term titles in the data field are all messed up and all over the place. So it's kind of hard to even know, what you're even going for. but have you had Luke Bruce on this podcast?
Louis-Francois Bouchard: Nope. I didn't that.
Avery Smith: Okay. I would highly suggest having Luke Barousse on the podcast. He's probably the, the leading like AI for data analysis experts that I know. It makes a lot of really cool YouTube videos about how to use ChatGPT to analyze data. but one of the cool things that Luke's done. Is he's actually web scraped, about a half million different job listings for different data roles.
And then done some basic analytics showing you what [00:19:00] skills are listed the most often. So for like a data analyst job, it's typically SQL, excel and either Power BI or Tableau, maybe Python as well. Just depending on how the seniority of the position, but that's in my opinion, those, I don't know, four to five skills are basically the skills that you need to know.
I typically, I, I, out of all those skills, I like Python the most, but I tell people to learn it last because it's, it's. The longest learning curve out of all of those skills, Excel. Most of the time people know Power BI and Tableau. If you can figure out PowerPoint, you'll be able to figure out those because just click and drag.
And then, SQL, at least for data analytics, basic data analytics, it's really like 20 different statements slash commands that you can probably learn. In like two weeks. So, those that's where I tell people to start and then you hopefully can land your first job. And then I try to tell people you can actually, you know, get paid to learn Python on the job, because it's like in the data world, you're always going to be [00:20:00] learning companies know that.
So they can actually pay you to learn those things. That's where I would start.
Louis-Francois Bouchard: Hey, this is just a quick interruption to remind you to leave a like and a five star review depending where you're listening. If you're enjoying this episode, I'd also love to know your thoughts about education in general.
Will online education replace the traditional education with universities and graduate studies? Do you think a paper will still be relevant down the line? I'm really wondering on my end, I think online education will change a lot. evolve a lot and be more and more well seen by companies when it comes to hiring people.
I also think learning by building is the most essential thing you can do. So where you learn your skills isn't really important. What's important, it's what you've done with it. So please, if you're looking for online platforms to learn, focus on the platforms that teaches you through building real world like projects.
Now, let's get back to discussion.
I don't know if data analytics is different, but I know that In artificial intelligence in general, the field is. [00:21:00] Changing and evolving super quickly. So, for instance, universities are very behind in what they teach, like the frameworks, the libraries, et cetera, what you usually learn is not that relevant anymore when you graduate.
So I wonder if it's the same for you and for your platform and how you manage to keep the content and everything up to date for all new students, if it's. Like if the field is changing so fast,
Avery Smith: Yeah, that's an interesting, interesting question. I definitely think AI is probably moving faster than data at this point.
there's just so much stuff happening, especially in the generative space data, like obviously is, is changing quickly, but it's not. It's not in terms of months, it's more on the terms of years at this point. So like, for example, I think the big changes that have probably happened since I started my program, what two and a half years ago.
So like, I'm, I'm still not like all that I've had decades, you know, teaching. Right. So I started this program about two and a [00:22:00] half years ago. One of the things that happened was, Excel came out with the X lookup instead of the V lookup. It just makes looking up things in Excel way easier. So you just make a new lesson about, about the X lookup.
You know, that's, that's fine. Python for Excel came out recently within the last year. Personally, I don't understand that yet and I don't really see it. Being used in industry. So I haven't really felt any pressure, although I probably will in the next year or so create some content around that. I haven't really seen it being used.
So it's like, I'm not going to teach something if it's not useful to industry. so yeah. And then like power BI camera. Power BI came out probably 2015 or something like that. so like, I don't know, maybe once a decade, a really big tool will come out that that's fairly new. but I don't think we've had, I mean, probably the newest one to come out for that is, is ChatGPT and generative AI.
But the problem is one, there's not really a league leader for [00:23:00] You know, analyzing data with, with generative AI yet. I mean, you probably ChatGPT is the front runner. but it's difficult because you run into a bunch of privacy issues, right? Because no company is really going to let you upload their datasets to ChatGPT without the proper protection.
So it's not really being used that much in industry, at least with the, with the proper. ChatGPT tool, you know, some of the open AI stuff that might be used and, you know, custom LLMs and stuff like that. but it's not being used that much yet. but it's still really interesting to potentially use for troubleshooting.
So we have a couple of lessons on like how to troubleshoot with ChatGPT and stuff like that. So I don't know. It's, it's a really good question. I'll, I'll ask me in five more years and I'll have a better answer.
Louis-Francois Bouchard: Yeah. ChatGPT is definitely useful for debugging. At least if you cannot put like sensitive information, that's, yeah, there's always. There's tons of things you can do without putting sensitive information. And speaking of chatgpt, how do you personally use [00:24:00] AI on your platform? Like either for building the content or for whatever it is being used to do on the platform, like not, not teaching the students how to use AI, but how are you using it for the platform?
Avery Smith: Yeah. So this is, this is actually something we just implemented. that's, that's brand new. So one of the things we did is we actually took all of our lessons, all of our video lessons, all of our text lessons, and we create our own chat box. And that basically is trained on all of all of the lessons. So it's like a custom trained chat bot specifically for my bootcamp, the data analytics accelerator.
And so it's just sits in the bottom right hand corner. Anytime someone has a question, they can just go to the chat bot, ask it there. It, it. I mean, I trained it basically, with some software. I'm not like that good of a programmer. So I use some software to do it. And it's like a little sketchy on like it's, it's responses sometimes, but, but it always gives references to the actual individual lessons, which [00:25:00] is super useful.
So even if it's not that coherent of an answer that it replies back to you, it'll often give you the references where you can go back to where our instructors are actually teaching it and, and get it straight from the instructor. So at least it does that pretty consistently.
Louis-Francois Bouchard: That, that's very cool.
That's, we also recently, well, a few months ago, we released our AI tutor with basically all the articles and courses we've created with TowardsAI, so pretty similar. And, and it's, it's been super useful and people love it. Is there any other way you are using leveraging AI on the platform? Because of course a chatbot is like, Really cool.
And everyone loves to just have a quick answer instead of talking to someone. But that's something like lots of people do. Is there anything you, you are using AI to do that is different?
Avery Smith: Yeah, I think there's some other cool things that we're doing, at my company that I don't see a lot, a whole lot of other people do it.
So, one of the things we'll talk about, it's another chat bot, but, but hold on, it's cool. I think you'll [00:26:00] enjoy it. So that chap I just told you about is for my students in the bootcamp. That's the purpose of it. I actually, when, when open AI announced the GPT Store, I was super excited. I was like, Oh my gosh, I'm going to build what I call AveryGPT, but it has like all of my knowledge.
You know, I'm going to put it in the, the open AI GPT store and it's going to be awesome. But then I realized in order to use the GPT store, you have to have ChatGPT plus, which is like 20 a month. And I was like, okay, I know a lot of people have it. I use, you know, the plus, but a lot of people don't, and they probably can't afford it.
So I was kind of discouraged when I was making it. I was like, ah, I don't really know if I want to make this. And so what I actually did is I actually created this thing called, AveryGPT. Maybe we can have a link in the show description or something like that. Yeah, of course. Where it's open to the public and it has All like 2000 of my LinkedIn posts, it has the transcripts from my podcasts, all a hundred transcripts from my podcast.
And basically it's like [00:27:00] anything I've produced content wise, at least via LinkedIn and via podcast, it has all the knowledge to that. So once again, it's a really fun chat bot where you're getting. You know, custom, like how I would answer it if you ask me a question, right? Because I love doing one on one coaching.
but it's just super hard because I value my time. You know, it's like if I'm doing consulting, I charge like 300 bucks an hour. So if I'm doing one on one coaching, I have to charge someone like about the same, right? which is, which is. Maybe not the most expensive, but it's expensive for a lot of people.
And so this hopefully gives people a chance to have one on one coaching from me. That's, that's more affordable, obviously maybe not as good, but hopefully more affordable. So that's, that's one way that we're doing it is, is making like a more publicly available chatbot, which I think is, is kind of fun. and then the other way is we've actually created some software called, I call it the interview simulator.
This is less for, for teaching data and more about actually helping you land a job. and the way this [00:28:00] software works is I recorded a bunch of my videos of myself asking really important interview questions that you're going to get asked in an interview, usually behavioral questions with a few technical questions.
And so I, you watch a video of me asking it and then a recorder pops up on your screen and says, you know, Answer the question and you record yourself on the screen, answering the question, and then you press submit. And then it actually shows an example, like a, like a good answer of how you should answer this, but this, this, interview question.
And then we actually give them a bunch of AI feedback on how their answer went. We rank it zero to 10, I guess, one to 10. We give the pros, we give the cons of what they said. So trying to get people a little bit like more than just a mock interview, but like a little bit interview coaching as well, which it was just pretty fun.
Louis-Francois Bouchard: Yeah, that's really cool. It's a. I don't even know if it exists somewhere right now, but it's, I feel like it's a, it's an amazing thing to have, especially on a learning platform. It's just really cool to have feedback on [00:29:00] the questions that you will surely answer. Like what, what we often say is to just go on interviews, to practice, like you will fail and that's it.
It's not a big deal, but that allows you to practice without. The, the shame of, of failing an interview. So it's pretty cool.
Avery Smith: Yeah, a hundred percent. There are some versions of this that, that are popping up on the internet. all of them are, are a little bit different. one of the things I like about, about ours is we're the only video platform.
So there are like some audio platforms and we're also the only one that uses. A real human to ask you the questions. A lot of them will use AI with like an AI avatar. Those are like. 80 percent there, I think, but they just kind of look creepy still, like most of them, especially unless you get like a really good model.
A lot of these AI avatars are still like, kind of like a little robot. So we, we, we asked the question from a human you record via, via video, which is the most real, that's like how it will be [00:30:00] in the interview. Right. Although I think there are some platforms now that are interviewing with AI avatars, which I don't like, I rather have the humans do it, but maybe that's the future we're going to, I don't know.
Louis-Francois Bouchard: Yeah. What did you think on, of, channel one? I don't know if you've seen the. No, I haven't. What's this? It's, it was like a, an automated channel media with fully generated thing that came out like, well, it should came up. I think it was supposed to come out in January. Not sure, but it was announced in December with a big video release and it was super hyped.
Avery Smith: Yes, I, I, I heard the audio of it, but I haven't seen the video of it. that'll be super interesting. Once again, I think I mentioned this earlier. I don't know about you. I'm sure you're worlds above me and like ChatGPT prompts and AI stuff like this, but it's pretty rare. Although with the interview simulator, we don't, we don't actually edit anything now that I think about it.
It's pretty rare for me to just copy and paste, you know, Any sort of generative AI, [00:31:00] because it usually could use some human optimization. So I'd be, but, but maybe they did. Maybe channel one was like, maybe they did like perfected over time. Right. So I don't know. It's, it'll be interesting to see. I'm here's, here's the truth is it's going to get better over time and it's going to need less humans over time.
I'm sure of that.
Louis-Francois Bouchard: But do you think people will want to listen to avatars and like see news from not real people?
Avery Smith: If it's compelling enough, like for years, they've had, I don't know if you can, you can cut this out if you want to. but they have the, for years they've had those AI influencers. Right. Yeah.
And, and so it's like, as long as it's compelling, compelling enough, I don't think people really, really care. you know, if it's not compelling, if it's like crappy, you know, I think people just care about quality. I don't think they necessarily care about anything else other than quality. And right now most AI, you know, products or I guess results [00:32:00] are usually just.
Not quite human sounding or not quite the high enough quality I was working. Cause we, we pump out, like, like you said earlier, how the heck do you, you know, how do you post on LinkedIn every day? I was talking to my team and they were like, well, what if we just used, you know, AI to generate LinkedIn posts?
And I was like, there's no one, I can't name one successful content creator right now that is putting out AI. exclusively AI content. Sure. They're using it to brainstorm. Sure. They're using it to edit. Sure. They're using it, you know, for this title or this part of it, but it's not like there's any, maybe, you know, one, but I don't know any content creator right now.
That's like, like just copying and pasting a result and posting it. And it's going well.
Louis-Francois Bouchard: No, it, and like, as soon as you have a little experience with ChatGPT, So you can already, like, you just see that it's generated, like, even if it doesn't use the, I don't know, the Delvin and all the [00:33:00] terminology that it always uses.
Yeah. Like all the terms are so proper to ChatGPT. I don't even know. It's, it's supposed to be using what humans use the most, but it's like, it uses a vocabulary that nobody uses. It's a bit special, weird, but yeah. Maybe it's done. Like maybe OpenAI just managed, like, to do this for the users to know it's generated, like some kind of, of management thing.
But yeah, it's, it's like so easy to spot that, that a piece of content or something is generated. This is just, it's really bad. And I guess that people that do not speak English very well, or that are not used to using ChatGPT don't know it. So they sometimes copy paste. But if you've been using it for like a month or, or whatever, you will definitely like right away know that it's generated.
And I don't know if LinkedIn and Twitter and other platforms have done some things [00:34:00] to like downrate the generated posts, but they, they surely don't do well. Like I've seen some generated posts, like as soon as you see the first sentence, you already know. Yeah, but they are, they are all very like posts that don't do any reach.
So I really don't know. I also do not know anyone using, well, copy pasting from tragedy or something in directly into social media. And I guess that's for a reason, but I guess. Yeah, I don't know. It's very easy to spot that it's generated, and I guess it doesn't have enough personality. I, I really don't know.
I, we, we've tried at Towards the Eye, we've tried to do some kind of, tool, like a social co pilot for, well, mainly for me, for podcasters. Bloggers, but to basically take a blog and try to follow the style of the blogger. Blogger, oh, yeah. While creating a post that would extract like one or two cool insights from the post.
But still it's, [00:35:00] it's so much work afterward editing the, the post, the post, like, it, we, we prompt, we just tried so many things and it's still. We, we are not able to make it good, interesting, and valuable. Like it's just, it doesn't work, unfortunately.
Avery Smith: I'm with you. So that, that was one of the other things I wanted to mention.
One of the ways that I'm also using AI is for each one of my weekly podcast episodes, I do on the data career podcast. we, we added it into scripts, which recently added a bunch of AI. like we always start the podcast with a highlight, from the interview. And so I'll ask to Script AI now to identify five.
You know, key parts of this podcast that could potentially be cliffhangers or gripping or viral or captivating. and then I'll go listen to all five of them and then choose one, right? I'm not having it choose for me. It usually can get one good one out of five suggestions. So we're, we're using that.
And then I also use a couple other tools to take [00:36:00] in the recording and transcribe it. And then they'll, they'll generate things like LinkedIn posts. They'll generate things, like, Instagram captions, they'll post, they'll, they'll, they'll do my timestamps. They'll do my titles and stuff like that.
And once again, we're never, it's, it's always copy paste, edit human, human edits. So, it's probably, it probably saves us. I don't know, a significant amount of time, but we're also still putting time into it. It's like, it usually gets you like 50 percent of the way there. So it gives you a warm start where you can, you can start from not a blank piece of paper, but at least some helpful suggestions.
Louis-Francois Bouchard: Yeah. And for just any creative task, just like, for example, creating a new post on LinkedIn, it's just like, you said that it takes you to the, to 50 percent of the job done, but. It's, I feel like it's even more than that, just basically finding the ID of the post [00:37:00] and then having a draft that you then just edit and post is of course, it's like 50 percent of the work is technically done because it was written and you need to edit like half of it and whatever, but just finding the idea of the post and trying to format it correctly.
Is this is our, this is a good, like judge is good for, I know, organizing ideas, I guess, and like knowing what to mention. And then if you are not an expert, but like knowledgeable enough in the, in the field that you are posting on about, you can just quickly see what judge says is wrong or right. And just edit very quickly and make a super good post out of it.
So it's. Yeah, I feel like it's just a, it's really a game changer for, for, I don't know if it's creative tasks, but like for a task that requires you to come up with something new or something with like very little guidance, I guess it's yeah, it's incredible. [00:38:00] It's awesome. I agree. Are you also teaching the, like your students to better leverage ChatGPT or other AI tools?
Like, do you have any lessons on that?
Avery Smith: Yeah, inside the platform, we are doing some lessons with ChatGPT, less about actually like analyzing data, for example, a lot with the career stuff. So like, for instance, we talked about the interview simulator software that we have. They have access to that. I also teach them how to conduct, you can ask ChatGPT to conduct a mock interview with you via text.
Yeah. and that turns out good. So a lot of career stuff, a lot of resume stuff, like resume bullets, brainstorming, looking at your resume for typos, those types of things. When it comes to actually learning this, the actual technical part of data, not a ton, but the one thing we do mention a lot is troubleshooting, like basically ChatGPT for, for me and for my students is now.
The new Google for when you hit an error code, I would [00:39:00] love to know what stack overflows traffic has looked like over the last year. Cause I have to imagine, although stack overflow is still useful. I'm not saying it's not useful, but I'm imagining it's actually decreased quite a bit because, you know, instead of me Googling and just clicking the first link, which is always stack overflow.
I'm asking, ChatGPT, you know, why did I get this? Here's my code. You know, why did I get this air? And then it'll be like, Oh, you got this error because of this. And then it'll often just be like, here's the revised code. So you don't get the air and that's really impactful. So really troubleshooting is really big for us.
And that's what we're using ChatGPT and AI for. The other thing is like warm starts once again, for coding. So whenever I'm coding. Something now in Python is usually my, my language of choice. I will ask ChatGPT to take a stab out of it. Like, like go ahead and try ChatGPT. Try to try to create this app that I'm, that I'm building.
And, you know, once again, it's only getting about 40 percent there, but that's 40 percent that I didn't have to type, you know? So, it's really useful for like starting [00:40:00] code. Like I always hate starting code. Don't want to start from a blank slate. Right. So give me the start of a code that does this. And usually it does fairly well.
Louis-Francois Bouchard: And other than programming, you, you said that you, of course you teach, but you also all, you keep learning and you are always learning. So do you have any, first, are you leveraging ChatGPT to learn new stuff, but also do you have any habits or good practices to like, do you force yourself to keep learning?
Or is just like, you, you love reading and you love going on YouTube and like, what do you do to just keep learning, keep progressing, improving? Is there, is it, Yeah. Anything, if you have any tips, specific tips with ChatGPT or just in your life in general to, to make you more productive.
Avery Smith: To answer your first question.
I just want to tell a quick story. So I'm an entrepreneur. I love doing what I do. I love teaching. I also love tech and I love software. And so I also love social media too. so recently I had an idea. To [00:41:00] build a SAS software that would basically help people repurpose content on LinkedIn. And I was like, I want to build this for me.
And then I want to see if anyone else wants it, you know? And, I'm, I'm a, I'm not really a programmer. I'm a, I'm a chemical engineer who learned how to program in college and then got a, you know, master's degree in data science. So I can code in like Python. But like, I can't code a website, for example.
This is how I want it to look. this is, this is kind of what I wanted to select the layout to be. and I was able to build the front end. And, I was able to build the backend in Python. I know Python, but still ChatGPT wrote most of the backend for me, but getting them to talk to each other on my local computer, I couldn't, I couldn't get [00:42:00] past an error with ChatGPT.
So I ended up having to hire someone else, but like, that's something that I would never be able to do without ChatGPT or actually I could do, I guess it would just take me a lot longer to learn. so that's maybe one of the ways I'm learning is just by like experimenting with stuff like that. Which I think is really fun.
And then to answer your question about like, how do I learn in my life? I love, I have like, I don't have ADHD at all, but I say that I do because I basically go through every second of my day being stimulated by podcasts. So I listened to, I listened to like 60,000 minutes. Of pot. Is that the right number?
I'll have to check my spot. Is it hours or pod or minutes? I think it's minutes, 60,000 minutes of podcasts last year. so I'm always listening to a podcast. So I learned a lot through that. If I'm not watching, if I'm not on a podcast, my TikTok and my YouTube are very like learning based. And then the last thing I do is I make myself read a nonfiction book every day, 10 pages.
I just get 10 pages of nonfiction reading [00:43:00] every day. So that's probably the one habit that's actually like a habit where it's like, not just kind of like. Learning is not just kind of happening because I'm bored, but I'm actually trying to like learn a little bit every day by reading 10 pages.
Louis-Francois Bouchard: That's a good tip.
Like it's not a, I don't remember, Habits? I think it's just Habits, the name of the book, but like trying to, to start new habits, you have to, you basically should. try to do very easy steps. Like just if, if you want to start running, you just put on your shoes and you, you go out and then you can just go inside if you want, but just start by putting your shoes and not, not like have in mind to run a 10 kilometer run each time or whatever.
Like it's too ambitious. Just try to start with little steps. So I really liked the. The idea of the, like limiting yourself to 10 pages, not like an hour or, or even more per day. And that's all right now it's million dollar weekend that you are reading.
Avery Smith: Yeah. I've, I'm switching up. So, I, yeah, I've read atomic habits and I'm [00:44:00] looking at my bookcase here.
I've read atomic habits and tiny habits and they both kind of talk about that as well. So I really like that. I just finished this entrepreneur book called, I think it's called the E myth. And that was good. Yeah. so now I'm reading, yeah, million dollar weekend for one. And I'm kind of co reading this like self help book called the mountain is you.
so those are the kind of the two I alternate on based off of, more interested in like learning about like business versus like. Becoming a more mentally stable human.
Louis-Francois Bouchard: And would you recommend both?
Avery Smith: I mean, to be perfectly honest, I'm 10 pages into, into both right now. So, so, so far so good. No, no complaints.
but I just finished the e myth, which for anyone who's entrepreneurial, and once is thinking about starting a business, I found it very useful. so I enjoyed it.
Louis-Francois Bouchard: Awesome. oh yeah, I had one last question about data for you. in the, [00:45:00] I guess, in the data analytics space or just data space in general with ChatGPT and everyone transitioning, do you feel like the freelancing or just job world in general is saturated now?
Like, is it already too late to join in?
Avery Smith: So I, this is so fascinating. I thought. That when COVID happens, that like everyone would become a freelancer. I just like, I almost envisioned a future where literally. No one actually has like a real job and everyone's just a freelancer working from home, working on different projects for months at a time.
maybe multiple at the same time. I don't know. That's how I envisioned the future. So I actually bought like a ton of like freelance. stock like Fiverr and like other freelancing platforms and they've tanked. They've gotten so they've gotten down like, like so much. So my theory was off or at least I'm, I'm too early on my theory, I guess.
So that's kind of how I pictured the future of freelancing. [00:46:00] and I guess we're not there yet. in terms of like data jobs being oversaturated, I don't think so. I think they're only going to get more and more prevalent over the next decade or so. there's so many companies who are still so far behind in their data maturity.
and data has such a high ROI that I see data scientists, data engineer, data analyst roles continuing to rise over the next decade. I think it is tricky because a lot of people see layoffs at like big tech, like fang companies and stuff like that. But like fang companies. I feel like they way overhired during COVID.
And so I think it's just kind of a reaction to that. So I try to tell people to not look at those necessarily as, as the, as the shining examples of, of where the industry is at, because there's still lots of data jobs in companies that aren't tech at all, you know, in manufacturing, they need data analysts, pharmaceuticals need data analysts, healthcare needs data analysts.
Like there's so many different professions and big tech is just one of them.
Louis-Francois Bouchard: Awesome. That's great [00:47:00] news for All the learners are listening and for the freelance on my end, I think more and more people are going the freelance way, but I think it's, I don't know if it's easier, but I think the people just like me and you, for example, are building some kind of community and network.
And so they find their clients themselves, like they don't have to rely on Fiverr, Upwork and all the other platforms. So maybe that's why they tanked a bit, like people do go the freelance way, but they, they create a YouTube channel. They are active on Twitter. They, they have a good GitHub or whatever, like they can find clients.
In other ways, maybe that would be my theory.
Avery Smith: You're a hundred percent right. So that, that could definitely be the case. And this is the tricky part. I think about doing freelance. I have a lot of people. In fact, when I first started my program, I actually taught you how to become a freelancer and like start a consulting or like an agency or something like that.
[00:48:00] And I stopped teaching that. Because it's way too much work, to become one, if you're transitioning, like you should probably work in the fields before you become like a freelancer. But even with that, if you're interested in becoming a freelancer, you know, from whatever tech job you have, I mean, great, go for it, but just realize that 50 percent of your effort is probably now not your tech field.
It's now business. you have to figure out how to market yourself, how to do sales, how to set up a business bank account, how to, how to invoice people. And it's that book that we mentioned earlier, the E myth, it talks about when you start a small business, there's, there's three people. There's the tech.
Well, even if it's just you, there's three personalities, the technician, which is, you know, Yeah. Yeah. You being a tech person that you've done your whole life, the actual technical work, there's the manager, and then there's the owner of the business. And all three of those are very hard personalities to, to balance all three of those and actually balance the jobs of three different people.[00:49:00]
I think that like, you know, there are people out there who do. Who do the freelancing route and they enjoy the marketing or the business. And so they're able to get clients that way. And I think there's people who are like, crap, I don't actually want to get into sales. I don't want to get in the marketing.
I just want to, you know, do AI or do data science or do machine learning. And they're like, crap, what do I do? And even like the platforms like Upwork or Fiverr, even then, like. Where you're not necessarily, you don't have to go out and find your own opportunities. The opportunities can find you even then it's a ton of marketing, intro calls, direct messages.
So it's, it's about 50 percent business, 50%, whatever else you want to do, if you start off by yourself. Yeah.
Louis-Francois Bouchard: If you just want to code and learn, well, I guess you learn a lot by being a freelancer and entrepreneur, but you learn different things. Yeah. But if you just want to go into AI data and just.
Play with the models and learn, like a traditional job is definitely. More [00:50:00] interesting. Awesome. Is there, okay, you, you have your data, the, the platform, the, data carry jumpstart that people can look into. And I, I also seen, I I've seen that you also recommend checking out the podcast and newsletter for the students, but just anyone in general.
And I just wanted to ask you, what's the. Difference with all three, like who should listen to the podcast? Who should take the course who should follow the newsletter or yeah. How, how would you introduce each of them?
Avery Smith: Yeah. So the majority of my, you know, content and products over the last three years have been for people with non traditional.
Backgrounds who want to break into a data career. So people who are like, Oh, I am interested in this. I think I want to do it. Maybe not a hundred percent. Sure. Yeah. Or not sure like about a roadmap. That's pretty much the, the people I'm speaking, trying to speak to. So for all three of those things, that's, that's what I'd [00:51:00] recommend.
if you're, if you're interested. In just like career switches, that's where I'd recommend the interview simulator where we, we practice interviews and you can do some like practice interviews and stuff like that. But really most of my content is for people who are trying to break the data analytics world.
so if that's you, then, then come over data crew, jumpster. com and you'll be able to find everything there.
Louis-Francois Bouchard: Awesome. And there's also, there will be the AveryGPT link. In the description. Yes.
Avery Smith: Yeah. If you want to check out just like a fun, GPT link. yeah, check that out. That'll be great.
Louis-Francois Bouchard: Perfect. Thank you very much for your time.
And thank you for building this learning platform and especially the, the, the jobs section that is super useful and really cool. So thanks. Thanks a lot for the insights. Thanks for coming on the podcast.
Avery Smith: Yeah. Thank you for having me. I appreciate it.[00:52:00]