Keep Your Mind HUNGRY - Interview With Alex Zajac

If you have a real passion towards it, it can literally end with you coding at night, being at like 1:30 a.m. saying, "Oh, I'm not sleeping, but I like it. It's, you know, it >> in this video, I got to sit down with Alex the Jack, author of Hungry Minds. We went through his career journey and how he found a passion for coding." Now, Alex shares a lot of information online, in particular about artificial intelligence and how that can help us as software engineers. and he's got a really interesting process for how he goes and curates information so that he can go share it with his audience. One of the really cool things about this process is that it kind of forces Alex to go learn about the different things that he wants to be sharing. So, it's always keeping him up to date as much as possible as a software engineer. I think you're really going to enjoy this video. So, make sure to check out Alex on social media. Sit back, enjoy, and I'll see you next time. If you don't mind, do you want to give viewers a little bit of background for for who you are and how you got to where you're at? cuz I think your story is kind of interesting from last time we were chatting. >> Yeah. Yeah, sure. Uh really really happy to to be here. So I'm I'm Alex. Um some of my friends in tech call me Hungry Alex as well because of uh what I'm doing in in the space. Um but yeah, my background is pretty I would say interesting. Um I wasn't the kind of kid that was like 12 and coding or doing robotics or you know assembling some stuff or or building some stuff. I was not like that. I was playing card games, playing volleyball, doing a lot of sports, but I wasn't into computer science super early on, I would say. Um, and yeah, I think I I got to develop this love for like math and generally like solving problems a bit later on when I was in high school and and before university as well. And yeah, when I finished high school, I got my my my first like kind of level degree uh from from France where where I was born. And I basically had to choose a school where to go. Um lot of the teachers and people around me are pushing me towards engineering and I was like okay why not let's go. I also um was like interested into as I said into math. So I was like okay financial engineering what is it I I what is it about like what is the job like? And I started researching more into that and I found a school that was quite specialized in in financial engineering. Um, but you know how it is when you're young, you see kind of a a path, but you don't really know if it's if it's good, if um it will fit you. You just choose it and and and and try. >> Um, so I chose that path uh a school in in France in Paris uh where I'm from. And >> and it went pretty good. The first two years I there was absolutely no code. It was mostly like preparatory classes, a lot of math and physics that wink wink you don't really need after. But it was kind of like nice to know to >> yeah it was kind of nice to like um expand your knowledge on that and it's always like work that you bake in and even though you don't directly use it in the job later uh it's also nice to have it as a ground knowledge but yeah I TLDDR I I started the school without coding third year we started coding which is pretty pretty late in retrospective um and it was in C which as well in retrospective would not have been my favorite choice I mean I think it's great to start with an object-oriented uh language for sure. Um I wouldn't start coding with maybe Python or super high level languages because you don't learn a lot of stuff that are important. Um but we might as well start in Java and I I submitted that feedback to to the professors and I I think they incorporated it. Um but yeah, I started coding in in C. I had really no clue what was programming, what was a function, a variable, um logs, nothing. and I stumbled upon uh what are right now two of my best friends, Mark and and Nicholas um which were this kind of like um nerdy boys. I would say that at 14 were building something robotics uh stringling something with Wi-Fi or like like anything that that that was related to computer science and they kind of like uh helped me learn to program. Um I started really enjoying programming in C. Um I started like building small applications and of course like school projects and stuff like that. Um but yeah I I after that I really fell in love into coding and the the things got fasttracked along that way. We did an exchange in um California um for computer science and we did a lot of like C++ data structure and algorithm. was way lowlevel uh programming and it was super super cool to visit a new country and at the same time learn a lot of stuff and after that I think we we applied to like 170 internship positions uh all across the world probably and we all the three of us got an internship position in Madrid in Spain for Amazon. Um, so we we kind of pushed I think one had the the offer first and then we kind of pushed for us to to be together and and it happened and yes from from there on yeah it's like super singular but from there on we we we continued the internship got all return offers and we started as as full-time in Amazon around October November 2021 and yeah from from now it's been almost yeah two and a half three years that I'm in the company I've to SD2 uh working on mainly uh search and also a bit more now on personalization systems inside um inside Amazon and yeah it's been a super super exciting journey and I'm super excited to share a bit more of things I know and things from my path here with you. >> Yeah. No, that's awesome. It's uh it's super cool to hear that uh like I I love the stories when someone's able to say like uh you know I I wasn't born with a keyboard in my hands and like wanted to start coding right away. Um because I think for for a lot of us, sure that is the case. Like I I didn't start, you know, as soon as I was born. But uh I was always kind of into computers and stuff. So it seemed like an like it it was almost like a no-brainer just like go in that direction because you already like this stuff. Um but the more people I talk with, I just love hearing that someone was like it's not what I was into and then I basically stumbled upon it or was exposed to it and started to realize like, hey, this is really cool. Um, I I love too that um the the way that you were talking about that it's almost like like it became a passion like you didn't realize that you were into it until you got exposed. And you you said the phrase like fasttracked, right? >> And I feel like when you really find like something super interesting, it's almost like it doesn't it's not work anymore. It's not >> a challenge. And maybe maybe I shouldn't say it's not a challenge. It's just that like it doesn't matter how big the challenge is because you're like I just want to do this. Is that kind of how it felt as you were like discovering programming and stuff? >> Like to 100%. It it's like I feel like programming is is literally like a canvas you can build anything on top of. And I think the first time I realized that is is quite early when I was starting uh coding with C. But before we move on, this is just a reminder that I do have courses available on dome train if you want to level up in your C programming. If you head over to do train, you can see that I have a course bundle that has my getting started and deep dive courses on C. Between the two of these, that's 11 hours of programming in the C language, taking you from absolutely no programming experience to being able to build basic applications. You'll learn everything about variables, loops, a bit of async programming, and object-oriented programming as well. Make sure to check it out. uh we had this assignment which was a image processing in C or navigation to process images so like filters over overlapping layers um basic stuff I would say but quite interesting to implement because it's the first time you apply like something like data structures and algorithms something you can see an app you can actually run before it was like sample programs and stuff like that it was super early but still it was super interesting to build and I remember this one I could spend nights like fine-tuning the filters or effects that I would do. Um, and it was it was like six months after I learn after I learned coding. I remember I did like this kind of um negative transparency effect where you know this kind of like first image blends into the other with a transparency background and it was it was hard like the algorithm was probably third. I don't know. But um but it was working and and we of course we couldn't process 4K images but it was a start you know then performance comes after after you do the job. Yeah, this was super interesting and yeah, with regards to what you're saying, that's what I love about that. And I think um if you have a real passion towards it, it can literally end with you coding at night, being up like 1:30 a.m. saying, "Oh, I'm not sleeping, but I like it. It's, you know, and this is all like for for my personal projects as well. Like I sometimes I forget about time because I really enjoy it." Um yeah, I think that touches a great point on on on that part. Yeah. >> Cool. Oh, no. That's uh that's awesome. And I I have that same effect where if I'm really into a project, it's like I will look at the time and I go, "Oh, no." Like I like I forgot to eat lunch or I for like it's already like 7:00 p.m. I thought that I just started on this. Um so it still happens. I've been programming for 21 years and that still happens to me where I'm like, "Oh crap, like the time has escaped me." Um but it makes a big difference when you know when you enjoy stuff. I know uh not everyone will feel that fortunate all the time. Even if you love to program, like especially with work, there's going to be projects and stuff that come up where you're like, >> "This one's not it for me." Like it just, you know, it's not cap capturing your interest. Can't, you know, it's not going to be perfect like that. But, uh, so very cool though. It's, uh, it's, like I said, it's awesome to hear that you had that experience where you're like, "Yep, I love this stuff." >> Yeah. Um, >> and I can also provide another example on this one of we kind of lost track time and it was with exactly the same friends I I went to the exchange with and that I started kind of like coding with is um we were building an application in React Native. React Native was like I think super new now it's it's uh growing technology and we're like let's build this app because we have this problem uh um that we want to solve. Let's try to build it and learn from it. And oh my god, the amount of time we spent into that. Like the code for some components was spaghetti code. We were still starting to like to like build something with a front end, a back end, a mobile interface like and it was super enjoyable. Even though like we launched it on on the Play Store in the App Store, it got like a couple of thousand of downloads. Of course, our friends and close family were super hyped about it. Like every personal project you launch like this. Yeah. >> But it it was actually cool. We saw like some some logs from people uh in the south of France using it. Some people in the north and east of France using it. It was an app basically to schedule like who brings what at a party, right? It's always a problem to know I'm bringing one or two pack of beers or one iced tea, one coke and it you never know the exact amount. People are lazy to count. They forget and then you kind of like it's not a hard problem I would say but still it was something we wanted to solve at that age, right? And we solved it a bit with this um again in retrospective way more stuff to do better on the code side on the product side. But still it's it was great learning and again one of these examples where passion like trumps you know oh no I should just only work one hour and enjoy. No if you enjoy a bit more and you know it's it's a cool trade-off for you and it passions you then go for it. >> Yeah. So yeah, having friends to to build stuff with is also like super valuable because like again when when the motivation happens to dip, if it does, then you have friends that are there like also building and you're like, "Hey, like no, it's you know, you you can feed off their motivation and passion as well, which is great." Um, and I I like your example, too, because it's a good reminder. You're saying in retrospect, you know, code was spaghetti. There's way like way better stuff we could have been doing, but guess what? You shipped it. people were using it like >> yeah exactly >> it didn't have to be perfect right >> of course of course and the second like good I would say side of that story is that when you're in third or second or third year at least where we were it was really not common for people starting to program to launch an app that was it was working it was not like beautiful it was not I I I mean I emphasize a lot the UX the front end stuff because I still love that but it was working right and and I think that's the point when you come to an interview you didn't have any experience We build this app, there's 100 people using it. I can show you the logs right now on my computer. And this is something super powerful as well. When you start um when you don't have really experience, you have you are in this cold start problem of I need experience, but to get experience, I need an internship to get an internship. I need experience. You know, this cold and chicken problem. Honestly, the best way we found uh with my friends and I was was to build stuff uh show it even though it's not perfect and and just, you know, write about it. Write an article on Medium or a post on LinkedIn, write a tweet, uh write a thread, whatever. Uh just expose it even though it's not big. Uh it's something that at your level is big because you just started programming. You're a student. Nobody expects you to build the next Facebook or Instagram or whatever. >> Yeah. And that's like, thank you for mentioning that. I think that's uh critical advice because it comes up all the time, right? There's and it's understandable, especially people trying to break into the industry, more junior software engineers. It's like, what can I be doing like to to get hired to get that exposure? Um I I always recommend to people to be building stuff. And I want to add the caveat that like not everyone has all of the time in the world outside of their like, you know, whatever they're doing from their 9 to5 currently. I get it. But like because it is competitive, like other people are doing this. So if you want to be competitive, it's it's almost like I'm saying you don't have to, but it's it's like >> probably the thing you should be doing if you want to stay competitive. But um the other thing you said that's that's awesome is like >> the goal when you're doing this kind of stuff is not to say how do I how do I launch, you know, a billion dollar startup? Like that's not the goal. you need to be building stuff >> which is going to give you practice and even better if you can ship something like again I don't say that you have to ship something but if you can what's interesting about that is that you have some evidence that you're you're now supporting something and that's another part of when you're working it's not just I'm going to code something press commit and push and then you're like hey it's done forever like I never have to look at that again like >> no you're going to spend more time looking at that unfortunately Ely. So yeah, getting something to that state where you have users is super powerful, talking about it online, being able to just like communicate things to like for example, you wrote the code. Like when you go to write an article about it or to tweet about it, what you're not doing is just copying and pasting the code and saying like that's it. You're you're communicating different thoughts about it, right? So um so you had uh success doing that as well then being able to kind of talk about those concepts online. >> Yep. Yeah. Yeah. I mean I I remember when I I was at that moment I was doing a few posts on Medium on React Native some animation things. Again I I didn't publish like a 100 post but again it was something out there that I I was push I was publishing. I you know just this the feeling of pushing the publish button is quite powerful like it's mine. I did it. Even whatever the quality it is, of course, when when it has a high quality and you're starting, it's great because it shows that you have learning and it shows that you have growth into what you're doing, too. Um, but yeah, I I started writing on that. I kind of stopped like having these blog articles, but now I have uh something else that we can touch base on uh with my newsletter and my content creation on LinkedIn. Uh sharing things I learn about, sharing uh links I found interesting for engineers and future engineers. Um, so it's always about putting yourself out there. I think that's quite important. And I I also totally understand there's some people that might be more I would say introvert or extrovert, but there's always a way you can find. You don't have to, I don't know, show your faces. You don't have to um be recorded. You can just write. You can just uh do that through code through uh reviewing others people's code and build an influence on that. There's a lot of space for that. It doesn't have to be, oh, I need to get on camera and produce 10 shorts a day to be interesting. No, I don't think that is. And in any case, the attention span is something that is is so moldable that um at some point you have to find what works for you as well. There is no generic advice on yeah, I should write X articles plus X videos. It's too hard. You should it comes back to the first topic. It should be about passion and about putting yourself out there and also working towards the goal that you have. If your goal is to get hired for example, you don't have experience, what we talked about before, uh building something in showcasing it might be a solution. But again, there is so so many factors that that go into that too. >> Yeah, that's great. And I think you know good advice to because I I think that like I have recommended to people to build in public I and so and and learn in public. Uh I I recommend that people don't take it can be a tricky balance because I think some people hear that and they go okay well if I'm learning in public then I need to be like I think about it as an influencer but the problem with that is like generally that influencer mindset is more like I I am an expert in the space you know educating you on it but I think that's a mistake because you don't want to learn in public and say hey look I'm an expert because because you're not and that's okay but like >> that's fine >> being able to learn in public and saying like, "Here's this cool stuff I'm trying out. Here's the takeaways that I have." What people don't realize, I feel like a lot of the time, is that you don't need to be an expert to be able to have a meaningful impact on other people because there's other people >> doing the exact same thing or they're a step behind you and and they can look at what you're doing, go, "Oh, cool." Like, like I'm a I wanted to go learn React Native. and the fact that this other person was kind of stumbling through it and they had some takeaways like now I don't have to stumble through it the same way like that's super cool. Um, so I think it's great and I like your suggestion that if you want to start doing that kind of stuff, you don't have to. It's not like, okay, well, I guess I'm a content creator and I got to crank out a 100 videos and I I can't imagine being on camera. Like, if you're if you want to do that kind of stuff to start sharing, you need to find something that's comfortable that you'll do because otherwise you'll start and you'll go, I hate this, not for me. >> Yes. Exactly. It it has to be aligned with an incentive that is strong to you. The incentive is weak or doesn't hold. I found at least for myself that it it just doesn't stick. It just doesn't stick. Yeah, that's uh like 100% that. So, uh not a lot of people know this because I've only started from for my content creation. It's only been ramped up the last 18 months, but I started my content creation 11 years ago. >> Oh, wow. Uh but I I gave up because the incentive like you said the incentive wasn't strong enough. I wanted to do like a learning and public thing for uh individual contributors that have gone into engineering management and started blogging and stuff and I was like wait like this is a lot of work and no one's reading it or there's no like tangible benefit that I was seeing. >> In hindsight I should have just kept going but um you know it's like the the incentive wasn't strong enough and I gave up right. So, I think yeah, you got it's got to be aligned to you. So, that's cool. But you have a news. Let's talk a little bit about uh cuz I know we want to get into some AI topics. Maybe we can hear a little bit about your newsletter in general, what you like to talk about and then we can kind of go down a more specific path on some cool stuff that's going on. How's that sound? >> Yeah. Yeah, definitely. Thanks for for providing that. Um, so yeah, my newsletter, Hungry Minds, I I started writing it, I think a year, a bit more than a year. are celebrated the year at the at the beginning of June. Um but yeah I I basically started because I was following a lot of links uh newsletters on deep dives or tips tools or GitHub repos that I really love and I was reading about them and I was uh exploring them going into the rabbit hole as we say u but I was like if I'm reading all of this I might as well um build something out of it and try to launch it and make it work and also benefit other people because I have a very very specific view of the world, right? These newsletters, these repos, the things I'm tracking to read, but others might not have it because uh they're not um that much into the same thing I'm I'm watching or something something else. And so I started hungry minds as mostly a creation curational newsletter where I share I would say deep dives, link on engineering, on leadership, um GitHub repos, sometimes tools as well. Um, I also have every every week a coding tip uh and of course a meme to kickstart your week on on the right code. I send that on Mondays uh quite early for US time zone and um at noon in Europe and yeah, I've been getting quite a bit of traction with this and my content creation on on LinkedIn as well. Um I'm I'm planning a lot uh to add features to it. Uh maybe launch a second column with dedicated deep dives on certain topics. Um so I'm planning to expand data also maintain in the back end I would say a system to create my links to make sure I focus on the right things. Um and yeah I think it's a pretty fun right to you know build around the tools like substack and behive as well that I'm using for my email list and try to be creative into uh how to spread it to people how to collect feedback and how to develop it. Uh I think that's those are the three pillars right that are important when you're building a product. Um, so yeah, I I've been having a lot of fun doing it. It's uh it's a weekly one. So every every Sunday I kind of review and and and prepare it. And yeah, I've been having good good amount of feedback uh outside of my company and also inside of Amazon. A lot of people uh like reach out to me saying this is amazing. Uh please continue. And this this makes me so happy to see those kind of feedbacks. Uh because as you said, it's about incentive, right? incentive is not one metric that it's binary there or not there for me it's a multiple blend of things how much I enjoy how much it's cool to code with it how much people benefit how much uh audience you can create and and all of this stuff right I think it creates also a great platform for people to to you know read and gather around the fact that they're hungry to learn more and that's why I name it that way >> yeah that's awesome and it's I think you you mentioned something really cool there where as you were saying like you want to have alignment with the things you're doing. You want them uh you want to be passionate about it, but it doesn't have to come from one source, right? So sometimes like a really something that feels very intangible is like how much it's helping other people or maybe how much um like truly the impact is there. And to give you an example, cuz obviously in terms of metrics and numbers, like I think if someone were to look at what you're doing just on the surface, they go, "Oh, like there's a lot of interactions, like a lot of subscribers, a lot of followers." Like those feel like very uh quantitative. They're easy for even someone else to go, there literally must be some impact here just based on numbers. But something that we it's like harder to to gauge is like when you have people that will message you where like you'll run into them at the office or something and they're like, "Hey, I saw that thing that you did and like that was really helpful for me." >> Like it only takes a couple of those for you to be like, >> "Oh man, like people people benefit from this." >> So >> yes, >> that's really cool that you have that experience. >> I think that's and that's a metric that is invisible, right? it it's hidden from outsiders perspective. I I only can see it as a creator, right? And of course, when posts are public, you can see the comments and everything. So that's that's fine. For example, on Substack, but I think as you said, this is super valuable. Like um the other day, my director reached out saying, "Hey, this is super cool. Keep rocking." And we had an informal chat. Um people literally from everywhere in in Amazon, in the US, in India, in Europe reach out to me as well for that. uh looking for guidance and I try to help them as much as possible but as you said there is nothing like um I'm sending my newsletters as I said every Monday and every Monday like in the first two hours I see like two or two and a halfk people opened it in the first hours and when you think about it and just like step back two and a halfk of people is a lot of people it's not a stadium of of soccer it's not like you know uh something like that but it's a lot of people if you put them side to side opening what I what I write about so Um so yeah that's why I want to develop a bit more the the writing part maybe like have a second more writing focused rather than creational focus uh column. Uh but that takes time that takes a lot of work to integrate to make sure it's it's scalable and I can deliver it as well because it takes a lot of time to create links. Um but yeah I think that's that's great points that you raised. >> Yeah. No, that that sounds awesome. And I think obviously with content creation and the amount that you're doing, um I I see there's like a a benefit for yourself where you get to go like you're basically forced to go learning different things. Like you have a you now have a commitment because you have an audience where it's like people are expecting >> this newsletter from you. So um I have a couple of questions and I don't know the order to kind of approach them in. So, I'm going to put a couple in front of you and see uh maybe what you want to talk about, but one that I'm interested in is like uh balancing your time because obviously you have your job, how do you make time for this and like life uh outside of it? Um and the other thing that I want to ask about is like is the learning opportunity, right? So, because you have this commitment like um how is it shaping how you're learning and how you approach learning? So, those are two questions and I just wanted to feel that out. Yeah. No, definitely. Like I think again balancing time is something that is super personal and there is so many like guides and you know like um books on the on the subject and I think there's general advice that kind of works for for most people but for me what I what I find the most um effective is the time blocking technique. Some people are are have different uh different ways of managing your time and for me time blocking works really well. So in terms of balance um I really really optimize that metric in my life if we can call it that way. I think balance is super important. I really also like the the quote that um Jeff Bezos once said that was super controversial that uh work life balance uh doesn't really exist because it would imply that it's 50/50 and it's not true. I can have a balance that is I'm 70% work and 30% of my personal life and this is my balance or the opposite way around right and I think that's again coming to the same topic that it's it's quite personal and it has to feel right for you um as I said in terms of managing time I use time blocking my newsletter and my content creation I do every single thing on Sundays everything I schedule the post I schedule the newsletter I schedule and cross post things to Twitter Um, I will look around some links I want to share or comment about and save them in my notion. Uh, I use a super simple system as well for for my stuff. I use notion to take notes. I use Google Keep as a reminders. It's amazing. Super simple app. Works really great. And I block out some time on Sundays to write my newsletter and and write my content. This is how I protect I would say the time from the week so that in the week I can go three times to the gym. I can go see some friends on Friday. Um I can go um uh to work of course uh I can go do all this interesting stuff and I know I'm protected of this uh commitment as you said because I have Sundays. Now the problems becomes when I don't have my Sundays when for example I'm in vacation or people are visiting and I I can't um right. So in this uh situation I also have a SOP a standard operating procedure as I call them where I have a lighter newsletter and lighter content. In that um in that strategy I either reuse some of the post or I have a backlog of things saved that I know can can work quite well and I draft from them for my newsletter. I might share a bit less links, run my algorithm in I would say fast mode, review uh I would say more likely the articles. I still uh with the the exer, the title and everything, but I I if I have one hour, I have one hour so I have to squeeze it in. But I make sure that I do it on Sundays and it's always done for a bit more than a year right now. I'm super proud of of that consistency. I think it requires great effort and it's also great for um for the mind and it also as I said acts as a mechanism so that I'm never like stressed what I'm going to post tomorrow what I'm going to post in a newsletter and along the week I collect links with pocket pocket is a great tool to book my stuff and and get it later there's also read wise that is great but pocket is simple and works everywhere um and so yeah I come on Sunday also I'm protected a bit on from Saturdays if I go out on Saturday I know I cannot go to sleep super late because I have this other day to work on it. So it's a high commitment because people might say I actually work six days a week but in in that instance I work for me for the content and for the people I I try to to learn my insights from. So I'm fine with that trade. >> That's uh super cool. I think the the parallel I want to draw here too is like right now we're talking and because I asked you right we're talking about the content creation you're doing, how you balance it. Um and then for folks that are watching if they're again software engineers if you are again a lot of people are watching this that are trying to break into software engineering or maybe they want to switch uh companies or whatever and they're trying to say like how do I kind of level up and get ahead but something that you said that I think is super powerful uh well two things in particular the time blocking as a strategy because that will that will work very well for some people other people maybe it's not for them but it is a strategy I use it for some things and it works really well and other things I block the time and I go great I block the time but I'm using it for something else. Um so I have to be careful >> fixibility. Yeah. >> Yeah. Yeah. So I have to be careful with it but I think it's a great strategy. But the other thing you said that's a huge important takeaway for anyone not for content creation specifically but for anyone consistency. So, you were able to carve out time every Sunday like as much as possible, right? There's going to be things like vacation and stuff that come up and you had mitigations for those, but you said, "I'm doing this every Sunday." And now you have over a year of consistency. And I think some people don't realize like it's not so it's not it's it's simple. It's not easy, but it's simple, right? And if you just stay consistent, the amount that you can accomplish in a year is like it's incredible. >> And you can apply that for building your side projects and learning things. If you said, "Every week for one year, I'm going to try building a little project in some new tech stack," you would learn things in a year. >> Definitely. And the hardness comes from that. I think you're right. The hardness comes from the fact that it's um simple but not easy. there's weeks where you're like h I'm tired or you know had a bad week uh somewhere else and you don't want and that's why I have as I said this I would say standard operating procedures where it's it has to be not hard for me to do it because if not there's a high barrier right and you're like h no um in the first beginnings when I was doing hungry minds and and my newsletter it was taking me three hours and a half to do it every Sunday that's a lot of time it okay the Sunday has like 10 hours but still a lot of time. Now it takes me around an hour to do a newsletter. I'm trying to bring it down to 20 30 minutes with like custom systems. So I think that's also key. Like if if you have a goal you want to accomplish writing for a year, get X amount of subscribers, um get this uh amount of followers, uh learn about this amount of stuff, you have at some point to stay consistent. And of course, I think as as you grow older, there's way more stuff coming into your life at some point in a family like stuff like that. So it it's harder to maintain. But if you start with that mindset of I can allocate that time consistently, then I think it kind of grows on you and I really enjoy doing it today. There is not one day where I'm like, I hated this Sunday. It was awful. No, because I do I do the newsletter on the morning, then I do some content creation. I take some breaks and have a film at night. It's a great Sunday for me. Um, but yeah, I I think again this is highly personal. Like a lot of people outside of like content creation or as well outside of coding or programming would be like, "But I don't understand you don't enjoy on Sunday." And I'm like, "I I enjoy what I do. I learn a lot of stuff." And that goes to to your second question as well. Like I I try to prioritize from the things I share what I read in in in entirely and which ones I more scheme and kind of like save for later. So I have a small system for that. Of course, I have a queue of a thousand articles I still need to read, but I try to rank them somehow. Um, and and on terms of learning, that's why as well the the focusing on Sunday is quite important because then the whole rest of the week I can build stuff, I can learn stuff and I can do things outside of the programming world as well. >> That's super cool. And so yeah, I want to hear a little bit more about the the the focus around the topics and maybe and maybe we can pick a couple of them and kind of see how they're applicable for for software engineers, right? So if you're if you're going into like the curation process like how how much of it is exploratory cuz you mentioned you have a backlog of things. How much of it are you just like I'm going to go out and see like any area and just like what's the latest trendiest thing? I don't know how you approach it versus like I know that there's some very specific things that people want to hear about or that you want to hear about and get more information. How do you how do you balance those pieces? >> So this one is quite hard for me and that's why there's variety of like types of content being shown in newsletter. I have generally like um deep dives on big tech software design. So like how I don't know Shopify reduced by half of its latency, how uh Google managed to create this service like anything that would be like that. I have a lot as well of um career growth and uh leadership articles. I follow a lot of uh people on on Substack and I think there's great content out there. I could quote them but there's just so many. So you can try to see in my newsletters uh if you catch them. Um but I share content on that because I think it's key at some point. I'm now mid-level engineer at the company and this is where as well you need to like scale your growth not only with technical uh skills but also with other things. So that's where also I share it. It's basically things I like to read about or learn about. And in terms of uh how I balance it, it really depends on what what's out there in the week. I try to fetch stuff that is quite new. It doesn't mean that it's like flashier or or or like trendy or anything. It can be an article that is new but on an old topic and that's perfectly fine. Um but I try to you know I have a system that fetches like from Substack from other RSS feeds that um that I follow and I try every week to expand that system to see what can work to also kind of like mention creators I love collaborating with. It's also a win-win for for both of the parties. Um then GitHub repos I track track things as well. uh I track the news with again RSS feeds and I share more stuff here that is related to AI or developments or like um valuations of companies. So there's a lot of variety of stuff in in the newsletter and I intend to kind of keep it that way because I just like to learn and and see a bunch of different stuff. I love to start the week with a quote that will motivate me. I love to start the week with a meme that will make me laugh. So it's literally a bit of a projection of of me and so far I think people enjoy it. So, I'm trying to Yeah, I'm trying to also gather gather feedback and see what can be the next steps for for that. >> Awesome. No, that's that's cool. I love that it's like you said, it's a projection of you, right? It's it's nice that when you have something where you don't feel like you have to be a cookie cutter to make it successful, like you can say like this is this is my personal kind of take on things. So, that's that's awesome. Um, AI topics because I know this is a big thing for you. Uh, maybe a a general question to start this off. uh is like in terms of trends with AI and how uh like that that's affecting software development like maybe what are a couple of things that you're observing or like I don't I don't necessarily want to say put you on the spot for predictions but if you have some thoughts about the direction of how things are going like what's that look like to you? Yeah, I think AI is a great step forward when used, right? Um, what I'm saying by this is again it it's so easy to spin up chat GPT uh ask a question and copy paste the code and it might be working, might not be, but I think it it really has to be it's the same thing. I would I would par it with the same thing with social media. you can use it and get lost into it, get lose six hours on TikTok and be like, "Oh my god, it's 9:00 p.m. What did I do?" And and you can have the same parallel with AI. If you just blindly kind of like rely on it to provide answers without doing the thinking, you you you basically delegate the thinking to to AI entirely, then it it can be it can be a bit of a a situation because you you don't really learn or grow or, you know, apply things that you can learn. So I'm trying as much as possible to shift the problem around and when I fix something with AI um in in my personal stuff I try to understand why is it happening and and for that like there's a couple of tips that I I I I would love to recommend. One thing I do a lot is when I whenever I run into a bug or something that I don't understand, I first copy paste that part of the code or or anything and try to to ask the system to explain to me what is happening and why I don't why doesn't it work. Again, here I I speak mostly on I speak entirely on my personal projects because I'm not really allowed to speak about what what is in the company I'm working with. Uh but yeah um I am I am doing that a lot. I use of course GitHub copilot. I really advise people to try to use it or similar if you don't if you uh can't afford it. The free tools are also quite nice but copilot are really is really cool. U because it's literally helping you but I think the core principle is to always be the pilot and not be the co-pilot if if I can make that analogy. >> Yeah. You don't want to be the co-pilot. You want to have a co-pilot. >> Exactly. So that as as I said like there's different level of of difficulty of stuff and and for trivial stuff I can be super good at implementing it without any issues. You could not understand what I do and it would work in most cases but I think it's there's still a lot of value to understand why uh this is the case and it doesn't really slow you down. For me I found that uh it really helps me. Example, I'm doing a bit of uh development right now in um smart contract on Solana blockchain. This is in rust. Uh for me, Rust is hard. I think for all first learners, Rust is hard. And I just copy paste even dogs and say what is that? And I'm pretty sure in a couple of months there's going to be AI agents integrated into dogs so that people can ask about them and learn about it. I think this is great on the learning part. This is amazing because you don't have to Google, find a video, find a specific spot or not find anything in the docs. You can literally try to have that and and and be answered in seconds. Well, the agents that can >> interesting shift, right? The the shift that we're we're used to is I have a programming problem or any type I mean we're talking about software development, but I have this problem. You almost need to figure out like how to go find answers, right? like and some things you're stuck on. You're like, "How do I even search for what I'm stuck on? Like, I don't know how to go ask a search engine the things that I need because I don't even know how I'm stuck." Right? Like, it can get really complicated. A lot of the time we might have like, I don't know, you have code that you can like a little snippet you can put into a search engine or like an exception trace or something and we can get by. But now it's up to you to go sift through the things and basically navigate the interwebs to go find the right links to go find answers and you're kind of guessing at each spot. You go >> with AI and LLMs were able to kind of invert that to say like let me just describe the challenge I'm facing like in plain English. Let me describe it. And with the LLM's ability to go sifting through content and understand it for us, it just like replays all of the work that we would have been doing by ourselves. >> Yes, totally. And I I think as well like it's really great to to like practically I also really like generating like tests with AI and of course you need to double test check everything and again be the the main actor in that. But it's great because you can see how you can implement the use cases and also like uh protect your code like on edge cases and and all of that. And I I think as it's it all relies to the same principle that I really like. I think it was um it's a quote introduced by um um Mrs. Corzikov uh on on LinkedIn from Google uh that you never need to you never should delegate the thinking to AI but more the execution right tribal execution prompt that you customize and I do this a lot like I have a bunch of custom prompts to like either um I don't know u translate a list of links that are minified to something else. This is something you can do with code, but I have a lot of prompts that I I have saves as snippets so that I can quickly paste them in any LLM that I'm using. I love to use uh GPD4. Cloud 3.5 also just came out which is super good. Um I also use Perplexity as a as a I would say augmentation of of Google search. Um so I love to use the systems. I mentioned as well Copilot. There's so many that I'm experimenting with as well, but I think these are pretty consistent and I'm again we're taking the example of my newsletter. Uh I have a few prompts that I have saved that help me with uh with generating it and also like ideiating over stuff I can write. And I think the secret here is is it's super stupid but not a lot of people know about it. Uh as at least from the one I talked to is using snippet. snippet is basically a um list of uh I would say couple of uh a couple of layers that you would paste somewhere and you would you would paste a longer string for example if I have a prompt to I don't know summarize some code uh I can name that snippet sum type it on my keyboard and it will paste the whole prompt this is great when you do a lot of repetitive task or prompts that are the same and and for me it's the case I have like some prompt that are persistent and I use for for code or for content creation or newsletter and that's great that's a great mechanism there's a lot of snippet managers free not free I use Alfred or script kit uh which is also an an automation engine for for Mac and windows I think they launch for Windows um so yeah I I use those two tools as well >> awesome and so you mentioned uh I want to touch on the the testing part because like some more specific software engineering like techniques I guess so obviously testing is very important in software engineering uh there's a million ways you can test things and as soon as you start talking about testing on the internet you have everyone who's an absolute expert and only their way is correct but um with uh with testing and leveraging AI I thought you touched on it a little bit can we hear a little bit more about maybe we don't have to do like specific examples especially if uh I don't want you to talk about work obviously but um if you have a situation where you're like you have something that needs to be tested you have access to LLMs or co-pilot what what might be an approach that you can leverage to get the benefit of AI in your testing? >> Yeah, so there's couple of ways to to do that, but the the the most important uh I would say is providing as much context as you as you need for your task, right? Because if you have a class in in I don't know C that you're trying to test, you paste the whole class and your prompt is uh generate unit test for this class. It will be super generic. It would probably work with good models but it will be super generic. the the in that prompt the system has >> has no idea what are your use cases why are you doing that how much level of granularity you need on that right so I I found and I'm I'm trying as much to spread that knowledge too but there's probably people are way more skilled than that in but that than me but prompt engineering is a real thing the the quality of your prompt where it is for testing or for for other outputs of LMS is super important so I spend a bit bit of time like fine-tuning those and for a unit test for or even for integration test, it's important that the the system has the context on it. So you would literally start with something um like act as a professional engineer and software tester in that stack or in that language. Be quite specific that usually works for for me. Um then you would say uh okay I have this application we're trying to do XYZ as part of this I implemented this class that does A B and C and I'm trying to test it because XYZ again something else right and then uh you paste your code there and you're specific about what you want to to generate again when I try to um have a prompt that generates a lot of different stuff like for example if I ask to generate use cases then the So then to explain me it's three task in one prompt and that doesn't work the best. I I found that prompt de composition is quite interesting where you would have a lot of context on your task but the output is super simple. For example, as a first step you could say generate use cases that might be interesting to test with that piece of code given the application's context of what you're trying to build. This is great because you generate ideas in that it doesn't matter. You're still the main actor because you can decide like no that that doesn't make sense or oh this is something I didn't uh think about and after that you can replace those use cases and say given those use cases generate the code and again there is one specific task for the prompt and also use cases that you fine-tune and generate it with with AI or as well from any of your product like um specifications. So step by step uh single objective decompose prompts I think is a great technique. It takes more time and I think that's the barrier because um I've been seeing a bunch of friends being like yeah I paste the code in a prompt and it's done. You can do that but it it will be less specific and usually less good of a result. >> Very interesting. Right. So a couple of things that I'm taking away from that are um you can you can basically get things to work. In this case, we're talking about having like, you know, being able to write tests over a class, let's say. You can have it work with minimal effort, but the the value of that can be dramatically different depending on how you prompt. So, in this case, right, you do the very basics. Here's a here's a class. >> Hey, chat GPT, write tests for it. And like you said, will probably work. It will give you tests and great. So, it depends. Like if you're if your goal was just to to get a code coverage number to go up, maybe that will do it for you. But I'm you can tell that I'm kind of being like maybe that's not really a great motivation for it. But um so okay, we can get more specific. And >> I I like that you were talking through like um why like telling the LLM why you're doing this or like what what some of your goals are, not just giving it code and saying do tests. Um, I want I want to say this out loud because I don't want to forget to ask this question. I'm always curious when you tell the LLM to hey, you you are a professional X, um, like I want to come back to this and say like I want to see your opinion on how that influences uh, output. But the before I kind of jump to that, the thing that I wanted to come back to was decomposing the prompt. So, I don't have a ton of experience with with prompts. I would say like a very like I use uh chat GPT and LLM for things but um I would say that my experience level is certainly quite junior. I've noticed though when you stack too much like >> like context like having a lot of context can be great but when you ask it to do too many things >> I find that the value of that is like very diminished. So, I think you made a great point like split those things apart. Have them be very purposeful >> and you can still be uh you can be the pilot, not the co-pilot as you're going through those and saying, "Okay, the output from the first step, it listed 10 things. I like seven of them." So, like your input is is taking that into the next part. So, I love that >> and it's already refined, right? because you take those seven, you validate them. You can generate ids based on that yourself and it's already much more dense than just you know the shallow generated then write code and that's it. So yeah, >> but it took it took more steps, right? So some people go, "Oh, like I I can't be bothered to go spend an extra two minutes." >> Exactly. >> Those things are are valuable. Yeah. >> Yeah. Okay. So, no, that's really cool. I think that's super helpful advice. And the question I had to say it out loud so I wouldn't forget it is um when it comes to prompting because I've heard this before and I I don't know if I personally observed and I'm not saying it doesn't work but I don't think I've observed the effect of doing it but when you say chat GPT or whatever LLM you're using you are a professional car developer um and you have years of experience like giving it that context uh from your experiences how does that shape its ability to give you better output cuz like couldn't it just couldn't it just infer that it is an expert in C like I don't know like and I I just curious >> yeah definitely so I I I observe that this um was a much uh had a lot a much larger effect before I would say when literally chat GBT came out and all these models starting spinning up clothes just started as well with an anthropic from no clothes from anthropic um so it it was quite powerful there uh I as as time progresses, you're right, this is uh a bit less relevant, but I still find that there's a huge difference if I'm trying to um summarize uh a section, for example, on a blog post or something like that that I I would use for a newsletter. there is a a way I would say denser uh output if I specify clearly what I want the like the persona to be like if I say uh you're an engineer summarize this article so that people can understand it it will do a pretty good job with good models this is why I was saying as time progresses and models gets better it's still fine but if I say act professional copyriter that has seven years of experience in Java though put is a bit better and as Well, it would be oriented towards something, right? And when I it also goes with the context that I add to this persona. If I say uh make sure you stay technical and make sure you don't like uh over abstract things and and simplify things too much while just also summarizing. This kind of works well if there's a a better persona defined. Again, there's a bunch of like um different I would say templates for prompts like persona um action and like formatting. There's way others. There's prompt engineering.guide that is super cool. Uh there's a bunch of other resources that I share as well in in my newsletter. I I don't have all the names exactly, but I know I shared them in the past. Um and and yeah, those things quite work. It's the same with image prompting. There's a huge difference if you learn just a bit of the formatting, just a bit of how these models are conditioned rather than just saying do a rocket uh with an engineer on it to do like a learning resource post or something like that. Um, so I think that's that's also key for image prompting, but also for textual and and prompting. >> That's I don't know if I can summarize it back accurately, but I think I have a better understanding. And the way that I'm thinking about the the persona part, it's not it's it's almost like how it's like the lens that it's going to use. Like I think the summarization one is a very good example because you could take um you could take the same bit of content. So let's say that you have an article or whatever a small post or something that's the context and if you say like in the prompt like this is the persona that you have. How it wants to communicate a summary is probably through the lens of that person. So if you said you are a professional engineer that has you know 50 years of experience in these fields like it will pro I'm assuming will probably try to be extremely technical in in its summary. But if you were to say um just to give you another example, you are a an the same article but you're like an expert like social media uh content creator and you're not talking so much about being extremely technical then it might find a way to summarize in a way that's a lot more consumable for for a general audience. >> Exactly. Yes. Totally. Definitely. Yeah. And again, you don't have to exactly specify that in a person. As you said, you can like add that layer saying stay technical, use simple words or stuff like that, but defining a personal up top worked in the past and I think it's still quite embedded into into the the latest like generations of the LLMs. So, for me, it works quite well and quite better when I add one that is quite precise. Again, it's a one sentence. It's quite easy. You just have to think a bit through it. >> Interesting. No, I I think that's super cool. Um, I want to ask maybe one more question and uh I don't know if you have an an answer to this off the top of your head that we can dive into, but so you mentioned testing is one of the things that software engineers can use for uh use AI with for a bit more of a an advantage there. Um, is there another thing outside of testing that you would say is like another really big value ad as a software engineer? You're like, I'm a software engineer. I want to lean a little bit more into AI so Devon AI doesn't come and take my job. um like what what is it that maybe like something other than testing that uh people can leverage AI for in software development? >> Yeah, there there is just so many ways to do it and to do it right. As well as I said we we mentioned testing um core I would say development like adding code to codebase and and kind of like leveraging AI to to do something can be very beneficial again if you are always the pilot and still still keep learning about what you're adding. It's the same like I would say principle like when you add something in the past with stock overflow you just don't copy paste it and then call it gray. You need to understand it. It's the same thing right now. >> Some of us did but yes but but it's the same thing right and and I I think um outside of like core development testing um communication uh summarization and understanding in in in your projects is super important. uh if you're working through I don't know an issue on GitHub that has 50 comments the status is not clear um extract the thread summarize it done in 2 minutes you know where is right >> so there's >> that's a >> it's a good use case it's a very clear use case >> it's and it and it has value and there is not really a a drawback right the system can miss things but again the simple prompt that you can um you can store once use everywhere where uh would work. Again, if things are different, like if it's a GitHub issue, it's not the same as if it's an article or something else, but you can have you create one prom, you take five 10 minutes for 10 different things, a GitHub issue, um feature to implement u um an article, um literally I don't know list of links like you can create prompts for these things and just then reuse them. So >> there is this then uh anything that comes into learning as we mentioned a bit like I love I love I love that for learning um I was saying about rust for example I'm learning on that uh I can like open up the book of rust which is a great resource uh watch a YouTube video and then be like okay about rust I know x y and z don't know please explain to me why is uh borrowing of memory in rust why do we need it why does it work and it will explain it to Right? You can double check it, but it's something that if you were to understand it, you would need to search through the docs. Uh you would need to go through YouTube and find the right video. It's doable as well. And sometimes it's more reliable, but it takes more time. You can cross cross um I would say you can jump that that part. Uh try to have it with AI and then fact check it. This is usually faster. Um I found it faster at least. So I think for learning it's it's really great. I want to build um multi-threaded application in Go. No idea how to do that. You can have um a prompt for that specifi specified for like how to build or skip a project and ask for a couple of links when when the LM is uh browser enabled. You can ask for a couple of links to start with. Those things might be outdated. It's fine. it it doesn't have to be a perfect answer but at least you have a starting point and some code snippets you can start or how to kickstart a project or stuff like that so I I found that for learning it's also really great so I would summarize it to this like for learning perspective learning something new or going deeper into something that you love uh AI is great um for core development or test development use right again as we were saying um it can be really cool um for um documenting, summarizing, and also getting to know the context of something that is big. It's also quite great um to create content. It's also quite great. Again, huge huge disclaimer. Never be the co-pilot. You still have to own the idea, the the narrative behind what you're posting, but it can help you format stuff. It can help you uh find ideas to post about. But yeah, writing literally a post which at GPTM probably every content creator tried is not working. You have to um either delegate the ID creation, you have to either have parts of it being research or help you research for the post, but it cannot be the output is literally the post how you prompt it. It's it's just not working well, >> right? Yeah. You'll see it's pretty it's actually very obvious. Maybe some people don't know because they don't pay much attention to it, but it's I don't know how you don't notice this stuff anymore, but you'll see like in in the vast realm of whatever we're going to dive into and do this judiciously and it's like these are all words that people just don't see. So, >> yeah, exactly. >> No, that's that's super cool. So, Oh, go ahead. >> Yeah. Yeah. I was saying like jumping on that point. Sometimes you you see posts like this and you don't want to comment like did you do this because you know you didn't but it's true that it's something to pay to pay attention about. And I think it's it's the whole core like click moment with AI. It's uh it's not there to replace us. It's there to make us way way better. And I I really think and believe it will happen. Uh there's a lot of like you know like controversy about will programmers still exist. I I still think we will exist. It will will be maybe a different job name. Uh, I don't know. But but I I do think there is value into pursuing that and and getting better with with the use of these tools, getting used to to them because it can really kickstart stuff um for for I would say simpler task or even task that we talked about. >> I love that. Yeah, I've kind of taken the same philosophy where it's like a I don't think that software engineers just disappear like we're obsolete, but I think that the the things that we do and spend our time on like they will shift and uh I think the only way you become obsolete is when the you're ignoring the tools and the trends that are coming along like AI is not going away. So if you're like I just don't like it. I don't want to use it. I would say like you're probably putting yourself at risk because everyone else will start using it. They will leverage the advantages and it's not that AI just takes over software engineers. It's that the software engineers that are leveraging it will become better software engineers and have that advantage. So >> y >> that's very cool. Um well thank you for sharing a whole bunch of different ideas for how people can leverage especially software engineers can leverage different AI things like different concepts whether it's testing building out new stuff learning I think is such like an underrated thing like you know just summarize this in a way that I can understand it now all of a sudden you have valuable content that you can interpret like tons of awesome stuff there. Um I'll ask offline for links from you, but if you want to tell the audience like where can people find you? Uh they want to engage with you like what's here all the the places. >> Yeah. Yeah. I mean definitely for for me I think the most important points are on LinkedIn. So you Google my name. I mean you LinkedIn my name and you should be able to uh to find me. And then I have my newsletters. As I was saying it's available on Substack. So you can search for hungry minds on Substack or go as well to hungryminds. Um it's sourcing the the same like post that I publish and I also post on Twitter but I'm a bit bit less active. You can also try to see me there and yeah I think that's uh mostly it. >> Cool. Yeah, I'll definitely make sure I double check and get links from you after and I'll have them in the video description and comments and stuff. So Alex, thank you. This was super cool. I appreciate your time. I appreciate your insights. Um, and yeah, I'd love to catch up with you at some point in the future and see how the different things in AI are evolving for you and your experiences using them. So, thanks once again. >> Definitely has been a pleasure. Thanks a lot, Nick.