WEBVTT 0:00:00.000 --> 0:00:21.240 Programming Throwdown Episode 175: Resume Writing. Take it away, Patrick. 0:00:21.240 --> 0:00:44.395 We—I don't know that we've ever talked about it before, but probably have. We've talked about all the things; I every so often it ebbs and flows. I have a DSLR, and I had recently upgraded to a slightly newer one, but smaller than my old setup. So you know, this is like a physical camera. It's large, has interchangeable lenses. Um, now what is— 0:00:44.395 --> 0:00:46.724 The SLR is the digital? Oh. 0:00:46.724 --> 0:02:07.977 So D is digital. SLR is Single Lens Reflex. So way back in the day, if you ever see like old cameras, there's like two lenses one above each other. Those were reflex cameras, in other words. You could look through, and the lenses moved in synchronized fashion. But when you take the picture, the bottom lens opens up and exposes the film, but the top lens is what you're focusing and composing the picture through. And so these are—I think they were called them TLR, two-lens reflex. Um, okay. And then so they move to Single Lens Reflex, which has a mirror at a 45-degree angle behind the lens. So the light shines through the lens, bounces up into a pentaprism, and in through the little cup at the top that you would look through. Um, and then when you go to take the picture, the mirror drops down and exposes the film. So then they had digital DSLRs where instead of exposing the film, it exposes a silicon sensor—CMOS or CCD. Then actually now it's a bit of a misnomer. Most of the lens reflexes, the mirrors popping down are gone, and they're all mirrorless cameras. And so your the lens is directly exposing the sensor pretty much all the time, and then the thing on the back or the thing in the eyepiece is you know rendering the—basically it's like a monitor for the sensor. Um, okay. But I don't know; I still call them DSLRs because yeah, whatever. Yeah, that makes sense. 0:02:09.124 --> 0:02:10.035 At it, but yeah. So— 0:02:10.035 --> 0:02:10.677 The—so they give— 0:02:11.039 --> 0:02:13.444 better quality than like your phone. That's— 0:02:13.444 --> 0:04:09.409 Ah, so I mean that's the—that's the question, right? Um, and so my phone does pretty—I have a pretty recent iPhone, personally, but you know, I think most of the phones, you know, we've been in a bit of an opportunity for, I think, features to equalize a bit. Um, and so I have a pretty nice phone. It has a lot of nice features, takes great pictures if you take time to, you know, compose them properly, make sure they're exposed properly. Mine has an option to, you know, do them as the raw images rather than the JPEG images, so you get more dynamic range in the colors. It uses extra bits um to store the color information—so 10 bits instead of normally 8 bits. Um, so you can do those things, and it actually is pretty nice. But if you think about it like the just in your head, like the size of those lenses, the size of the chip that does it, however many megapixels they are, they're still physically very, very, very small. Um, and so they deal with a lot of noise issues, and they make up for it by they sell so many of them, the amount of dollars they can spend to, you know, do tons of software processing and have the best sensors and do all that is is a huge factor rather than, you know, these mirrorless cameras maybe they sell, you know, one tenth, one hundredth as many as they sell, you know, whatever the top-line phone is, and so they can't spend as much. But the physical size of the sensor, like you mentioned, can be a lot bigger. So on a—they call it like a full frame again going back, it used to be 35 millimeter film, which is 35 millimeters on the diagonal. Um, if you have a full-frame sensor, which is the one I happen to have, it's 35 millimeters on the diagonal. I mean, that's size that is like a huge percentage of my phone's total dimensions. Like that's a big sensor. Um, and so when you go to take pictures, if you just look at them, you know, in small size, they—they kind of they feel equivalent. Like if you just pick it up and snap it like you would with your phone, you— 0:04:09.932 --> 0:04:54.904 you don't really notice a difference. But again, if you take time to, you know, get a lens that has maybe a bigger zoom and, you know, that kind of stuff, you start to realize there's a lot of more data that you can pull out, there's a lot more color you can kind of shift around, you can kind of change the exposure because the camera doesn't—when you go to take a picture, you don't know what you're trying to show. And if you notice like if you go take a picture at sunset um and you pull up your phone or even one of these cameras and you take a picture, it's trying to adjust the white balance so you know you're not necessarily seeing the colors in it. You're looking out towards a very bright part of the sky with everything else dark, and so it's trying to average out across the whole scene. If you now in your phone, you probably have options, but people rarely use them. But yeah, I have no idea what— 0:04:54.904 --> 0:04:57.099 half those buttons do on the camera app. 0:04:57.099 --> 0:06:54.400 But when you have a physical camera, you have all these, you know, buttons and you sort of like switches and dials and all that stuff is sort of immediately there. And so for me it's a bit of a first-world problem, I guess, a luxury, I would say. But you know, you can do these things with the phone you have, but if you—if you're able to have one of these camera forces you into a slightly different way of thinking, and what you find is, or at least I do, that the actual process of taking these pictures sort of like treating them with care, which of course takes orders of magnitude longer than clicking the auto-adjust feature and posting them to, you know, family and in a group chat or you on social media or whatever. And I don't even really share these pictures with anyone, but just sort of having them and sort of taking time with the content picture. So I'll take a picture of my kid, but rather than just, 'Hey, smile,' snap it. It'd be like, 'Hang on, I'm gonna take a picture.' You know, and you have them sit, you know, a little nicer, and you get a little bit more of a composed thing, which is how growing up at least that's how kind of all the pictures were. No one just like rapid, you know, click, click, click, click. Like people would set up a picture; you would take a picture. Now you can say that's fake, but it has a certain nostalgia and specialness to it when you go back to edit that picture. You're sort of re-engaging a bit with that moment because you took time to kind of think through it. And then—and then it's a little bit of a more of a thoughtful process. So I've been finding that interesting is, and—to answer your question, I don't actually know because I treat them so differently. The force yourself to use your phone and to use it in the way where you adjust all the settings, and it's obviously not going to be as capable still like, you know, you can put a gigantic absolutely gigantic lens with a huge zoom, and your phone can't do that. But if you shoot in the sort of normal ranges of things, if you really take time, learned a sophisticated phone parameter app and you know figured all that, maybe—maybe they're the same, but just pointing out that it's kind of today our interaction with photography is just so far removed than even— 0:06:54.400 --> 0:07:32.685 30 years ago, 20 years ago, the crappy cell phone cameras like today. And then you know lots of news events happen, and you just realize so many people are taking pictures, taking video that even when events happen and people didn't realize they were going to happen, they end up getting captured. And it's just this very interesting thing that I didn't spend a lot of time thinking about, but I've recently been re-engaging with it and just thinking through like pictures as physical—like actually when was the last time? I don't know for me, it's been a really long time. Printed a picture out, like sent it off to be developed. Oh yeah, printed it or held it in my hand. It's an entirely different thing to like hold a picture in your hand and like look at it. It's— 0:07:32.685 --> 0:07:38.035 Crazy, remember the last time I held, you know, a picture? One of my own pictures. Dude, it's been—I 0:07:38.322 --> 0:07:53.475 Probably haven't had a picture developed until recently. I went on a trip and took a family portrait and had it printed for um, for my family. You know, I like a large. It was like the first time in probably five, six, seven years I'd had like a picture become a physical set of atoms. Okay? So, 0:07:53.678 --> 0:08:03.060 I have a couple of questions. First of all, like how do you learn how to take pictures? Like is this something you learned on the internet? Do you take like a—yeah, okay, got it. And 0:08:03.516 --> 0:08:13.320 I'm by no means claiming to be an artist about it. I just—it's an interesting process, and it's really—I mean, if you're an engineer, it looks—it's not that hard. That makes 0:08:13.320 --> 0:08:20.965 sense. So you went on a YouTube video and it's like, okay, here's what the different acronyms are, and here's, you know, when shutters one or 0:08:20.965 --> 0:08:28.120 the other focal length? Like you can learn these in an hour of, you know, watching YouTube videos and looking on, you know, tutorials. Yeah. 0:08:28.424 --> 0:08:42.329 Okay, my second question is like when you're doing something like printing a portrait, have you tried any of these like super-scale algorithms where like they'll like increase the DPI of your photo using AI? And did you need anything like that? 0:08:42.329 --> 0:09:38.843 So I did recently. I had a picture where I needed to zoom in a little bit more, and um, I actually did use which I think I could have done it on my phone too, but I happened to have used um my my large camera for it. I have a Sony mirrorless camera, and it has a bunch of extra resolution. So I zoomed in, and it was really noisy because you know when you blow a picture, it was noisy. So I actually didn't do super-scaling. I didn't really need more resolution, but I did run one of the AI like denoise of VisorDenoise of Fires um, and it did actually like—I don't want to say creepy because it didn't put more detail than was there, but it did a really good job. Like it was suitable. It went from like this is blurry and you can see what's happening to like, you know, it was a picture event for one of my kids, and so it was, you know, like them doing something up on a stage, and so it was far and it was dark. It was very noisy. And afterwards I was able to send it to people, and no one realized this was like a tiny corner of a picture that I took because, you know, wow, it's amazing. So 0:09:39.383 --> 0:10:40.780 it's amazing. Um, I recently used the thing where um you select an area. Basically, uh my wife had this photo and there were people in the background. Um it was at—it was at the beach, and unfortunately there was a woman in the background bending over, and so it was just like the most embarrassing shot of this poor stranger in our in our family photo. And um so, so yeah, there's this tool you just lasso that part of the image, press a button, and it paints it. And I was also pretty surprised. Um, you know, when you see examples, you're obviously seeing the best of something or, you know, someone's at least tried multiple times, and they're showing like when you read a research paper, they're showing you, you know, something that they're really proud of. And so I was a little skeptical going in, but it was great. I mean, it even drew like a bird. I was like, wow, it's like it had a ton of creativity. Um and uh yeah, it's just 0:10:41.027 --> 0:11:35.500 shocking where that industry, I think, but I think that's—to be clear—I think that's a relatively recent um development. I think that even I believe I've seen some press release from Google has been doing some of that um that they are offering that in their Google Photos service and stuff too. But I know Photoshop or Lightroom added some of the things you're talking about, some others, but I think it's been like a two-year kind of recent development. If you had tried this five years ago and you'd circled it, it would have done some sort of like color averaging blur texture stamping kind of thing. Yeah. Versus it wouldn't have like if you had a wave at a diagonal, it would have like cut a piece of a wave and put up a nut, but it wouldn't have lined up quite right. It would have left a thing. And it's always been possible for people to go in and do it; it just would have taken enormous amounts of skill and time. Now you just like I said, you just draw a loop around it, tell it you want it, you know, removed, and it will, you know, hypothesize about what was there based on what was in the input. Yeah. I 0:11:36.225 --> 0:11:41.676 mean it basically drew an even larger background, which is exactly what I wanted. That that word 0:11:42.975 --> 0:11:55.362 has become ambiguous in this circumstance because you did want—you did want behind what was behind the person. That's exactly—exactly all right. We're degenerating now. We better get on to don't be like where's the programming talk. 0:11:55.834 --> 0:12:00.339 Oh god. All right, news and links. Let's see. Patrick, you're first. 0:12:00.340 --> 0:13:41.033 Okay, so this one is an article on the blog by I should have looked at it, but the blog's name is Robert Heaton. So I'm going to go that—that's the author's name called Pi Sky Wi-Fi. Um and they don't give actually, as far as I could tell in the blog post, they didn't give a ton of specifics about what they were doing. But this story is so funny, but I wanted to use it as per usual to talk about kind of this—there's actually something really serious here. Um and what this person did is they kind of have a story. I believe it's probably apocryphal, but you know, they have this story about they were flying in an airplane and they wanted to use the internet, but you know, it charges you some amount of money that always feels like a little bit too expensive. But maybe—and so they were, they needed to do this work. They were going to go, but then they realized they basically could go into their account on the website of the company of the plane, and that was not considered like accessing the internet. So they could, you know, go onto the company's website, check flight times, go to their frequent flyer mile account, that kind of stuff. Then they realized within the frequent flyer mile there were a couple of their account settings that they could edit things like their name or, you know, you could imagine like your address. And then they had the broad idea of, 'Wait a minute, I could just start editing these fields and if I ran a program back home, I could, you know, basically start parlaying information by, you know, think like Base64 encoding.' There's—they used a different scheme, but you know, I could fill out one field as like a request for a website and have a server back home that could chunk up the website and put it into a different field that I could then read, and we could basically pass information at this super low rate. Right? You imagine? Wait, let me. 0:13:41.033 --> 0:14:14.732 Let me repeat this and see if I understand. So the idea is you know you can't get on the internet without paying whatever some nominal fee. Um but you can do for free is like, you know, in Delta or whatever, you can watch movies. There's like websites they do give you for free. Yep, you're saying one of the free websites is one where you can update, let's say your first and last name? Yes, on the Delta account. And so you're saying this person made his first name like google.com and then his last name is like the google.com website, like another server is also editing the name? Yeah. 0:14:15.188 --> 0:16:15.120 Like imagine it took the whole Google website encoded in Base64 and put it in his last name. Um that would of course make it really easy. There are of course character limits, so you also need to chunk it up, which means you need a protocol, which means you're probably not doing it by hand, so you know how to script and then a proxy sitting on top of the script. Um horrible idea. Very likely if they figure out what you're doing, you could easily—it's like tied to your account, right? You know, they anyways not not a good idea. I would not definitely not go ahead and don't try at home. A hundred different reasons. Um but the so two things. So the blog article is super useful because it goes into this that if you sort of the beauty of how we have the internet today is it's all these layers, right? And we tend to forget about them. You might learn them in school, but you know all the layers. There's our transport layer and then application layer. And—but this is the beauty of it is like if he can find a way to pass bytes reliably, you know, or even unreliably, but pass bytes from the airplane to a computer and from the computer back through some protocol no matter how slow, no matter how bad the data rate is, you can basically wrap that up, and then everything above it just doesn't care. Um you know, it may be slow like you go to google.com and you wait 20 minutes before it loads, but it doesn't matter. Like it'll work, right? You know. And so he this sort of thing. So one that's an interesting discussion, a very weird way to get into that discussion about the internet and that you know protocols behind it and that kind of specifics that there's going on. But this is a broader thing is that, you know, to be honest, I've been in this boat before where you know I did development on microcontrollers or something else really small. They didn't have even maybe a serial port that we could use in a way that would work. And you can find if you can find a way to turn something on and off, you know, we were turning an LED blinking an LED if you can do it and you have a way to sense it, you can. 0:16:15.120 --> 0:17:27.782 establish a communications protocol. And so we had a thing where like we would do a run, we would store up a bunch of data, and then the thing would go into a mode where it would basically blink out ones and zeros with you know dashes and dots like the telegraph, and we had a little like photocell that would record it, and then we would decode the zeros and ones, and we would get out the log messages. And sometimes I think we get into this day where we kind of forget about these things that not that long ago, you know, we were using telephones with audio sound to transmit the internet through, you know, like this is—this is the way it was. Um but it actually still true today, and there are times if you're going to do—I would call those things kind of hacker things—where you're trying to get something out where you don't have full access or not always definitely not in an illegal way, but just in a sort of surreptitious way. Uh it's just definitely a very interesting thought experiment to challenge yourself instead of just paying the definitely much easier, you know, whatever ten dollars, eight dollars to use the internet. Instead, you do, you know, it's horrible end around, but the point isn't—the point isn't like not paying for internet and getting this really slow thing. The idea is being creative and using the tools that are given to you. Yeah. 0:17:28.896 --> 0:17:55.120 Totally. One thing that just from a from an I guess like artistic standpoint, I really love the way that this person wrote kind of in the first person and kind of like toggled between telling a story and then going into some technical details. I feel like it's really well done. I love it ends with, 'I scrolled around the HTML and reflected that this had been the most and least productive flight of my.' 0:17:55.120 --> 0:18:12.906 Life. Um yeah, there's there's of course a problem there where if you kind of need to fly twice because the first time you kind of discover this, but you unless you pay for the internet, you can't actually access your computer back home or set something up on the other end. So that's. 0:18:13.277 --> 0:18:19.808 True. Yeah, you have to build it and pay for that. Yeah, that's totally right. But now he has a source code available, so. 0:18:21.310 --> 0:18:23.048 Anyone who wants to get banned. 0:18:23.419 --> 0:20:21.460 Yeah, that's right. Um all right, my new story. First new story is this video making anime character—sorry, animated characters with AI art. Um I thought this was so cool. I stumbled upon this because um I typed this into Google expecting to see a research paper on how to do animations. You know, like we have everyone knows about DALL-E and you know Stable Diffusion these things where you type in, you know, you know a horse eating hot dogs on the moon, and it'll like render that photo, right? But there's nothing, and there are things for video as well, right? But you don't have is something where it's like, you know, a 2D, you know rig of a horse, and what you end up with is this like this marionette puppet right that you can animate. Um and so I just thought this is like an interesting void here, like no one's built like a text-to-animation, um and searching for that, I found this video which I thought was just amazing. This is basically somebody who's they're heavily promoting Adobe products, although they're not it's not an official Adobe video. Um but what this person does is they start with text-to-image, and I think they put like a lion wearing a suit or something like that—a bipedal lion wearing a suit—and so you get this this lion standing on two legs, you know, it's pretty human-like figure, but definitely a lion head wearing a suit. And then this whole jungle background. Um and so he goes through like he deletes the background, but it's, you know, it's a—it's a complex high-frequency background. It's not like it's on a white sheet or something. So so he shows how you can use AI to like detect the background. 0:20:21.460 --> 0:21:47.640 in a sophisticated way, deletes the background. And then the lion, you know, because it's a photo one of the hands of the lion is covering his like midsection. So again using like lasso tools, but they're like intelligent, like it's kind of like snapping you to like the semantic segment, right? He like pulls the arm off, and then now you have this like hand-arm shaped void in the torso of the character. So he uses the AI in painting and it draws like a pocket on the jeans that was missing, you know, and all that. Um and now he has this arm that's kind of bent, so he unbends it using some other tools, and basically within about 15 minutes he goes from this image that was generated by an AI to this like 2D character that can blink and the mouth can move and everything. Um I thought that was really interesting, kind of like symbiosis of you know AI and real skilled artisan work on the computer. So it's definitely worth about 10-15 minutes. I mean if you're not super into it, you can kind of skip ahead, skip to different parts, but I was just enthralled with this person's ability and all the different AI tools that some of which I didn't even. 0:21:47.640 --> 0:22:55.679 Know existed. I've seen a few examples like like this where you mentioned it—a person with skill just becomes like much more powerful having access to tools and being willing to have a creative approach to the problem. Um So right, you know, being willing to say it doesn't matter that I have a hundred years or whatever—that's too many 20 years of experience drawing this scene from scratch. I'm going to see if there's an I'll call it better way. But also being willing to leverage the fact that I am still an artist, I still know how to rig a 2D animated character is something for instance I wouldn't know how to do, but they did. Um And so yeah, I think this is one of those—I'm curious to see what'll happen. Like maybe AI will just leap ahead and be able to do it from scratch, but much more likely in the midterm we're going to see willing people able to just scale their productivity unbelievably using approaches that are kind of blended or hybrid like this, and we don't really have a term to call it. Like it's not really AI art anymore, but right? Right that way. 0:22:55.679 --> 0:23:24.975 Yeah, that's right. I have a friend who maybe AI-assisted art or something. I have a friend who's in the games industry and he goes to a lot of these conventions, and he said last year, you know, if you went to GDC, one of these conventions, and you gave a talk on AI art, someone would just like stab you in the back, uh, in the heart, and drag you away—metaphorically speaking, like the audience would just destroy you with tomatoes. 0:23:24.975 --> 0:23:26.292 At you? Yeah. Yeah. There. 0:23:26.292 --> 0:23:58.840 Would be like no trace of you left on planet Earth. But but this year he went to—I don't remember where I saw GDC, but he just got back from another conference, and he said it's just everywhere now. He said, you know, that it is. It's kind of like he just—he described it as like COVID, where people tried to contain it, and then now it's just everywhere. So it's like there's whole companies with stands and booths, like, you know, talking about their AI tools, and people are giving talks on it, and there's just there's just no going back from that now. 0:24:04.530 --> 0:26:03.094 Wild times. Yeah. All right, my next article, I'll keep this short. The title is 'The Real 10x Developer Makes Their Whole Team Better,' and this is on the Stack Overflow blog. I just generally wanted to take a moment on 10x engineers. This blog post was—it's pretty short and mostly just talking about there's a sort of mythical 10x developer that floats around in various forms, and people debate, you know, are there really people 10x more productive? And the answer is in certain contexts probably, or even more. Sometimes people call them sort of rock star engineers or rock star developers, and the blog is trying to make the point that through a specific process—they were sort of saying, you know, communities of practice, I've heard them called centers of excellence, or people gathering together, and the real way to be effective is to basically help the team around you. And you really can 10x, you know, an experience or the ability of your team to its total output if you can really communicate knit together, encourage good practices, mentor—like these kinds of things. And it really does resonate with me. This is something that has become—I don't know, like a pet peeve is exactly the right word, but people, you know, I want to hire someone just like super, super smart, like just the smartest person in the room. And it's maybe—I mean when I look at them or think about composing teams that are working with such people, it's like, but how do they communicate? How do they work with others? If if they're super amazing in certain circumstances, it may be useful, but far more useful as a person that can, you know, work well with others, help, you know, make you know better than the sum of the pieces. I guess is the—is the kind of way of saying that. And it sounds goofy, but I really have run across people who are like that. They tend to be very giving, so they're not super, you know, self-focused. You know, they're always trying to look around and say how can they help people? How can they improve the stack? How can they make tooling better? Um, and I really do think there's the 0:26:04.410 --> 0:26:52.200 ability to just have a very outsized return on, you know, a single person if you consider it across their ability to help others. And in some ways, regardless of whether there are people who can be, you know, 10 times more effective code output, I think that the, you know, ability to knit together a team is sort of far more useful. And so when people think about their own careers, this is much less talked about—how to help others, how to make your team better. It's about, you know, we and even on this podcast somewhat. It's like specific skills you can learn, ways to make your own, you know, problem-solving better and more improved. And those things are important, but also thinking through, how are you a good team player? And it's a much more subtle and less easy to talk about problem. 0:26:53.618 --> 0:27:04.240 I think that's fair totally. Um, you know, yeah, I've thought about this a lot—the 10x engineer thing. I mean, one thing that comes to mind is like 0:27:05.802 --> 0:29:04.095 it's really kind of a bit of a faulty premise, right? Because if a person is a 10x engineer, then it's—it really just becomes about saving money. Like, like if a person is doing 10 times as much work, then there's nine—there's that's nine people you don't have to hire, and it just becomes this big money game. Like Cisco hires 10 engineers, Facebook hires one 10x engineer, and they both end up exactly the same. And I feel like it's kind of a false premise. I think that—I think you're right. I think there's several different things that you can combine with your engineering skill that are very like in high demand and low supply. So I think, you know, engineers that are really talented and can also kind of coordinate and influence a team is—is that's a really hard combination to find, and those people, I think, do really well. I think I think engineers that are extremely multidisciplinary and can build a prototype by themselves—um, that is also kind of rare. Like someone who is, you know, they talk about the mythical like full-stack engineer, right? But but really it's like, you know, you might be a front-end person, but you know maybe you have to train this AI because you have this idea that involves the latest AI and front-end. And so to be able—the ability to just go in and learn whatever it takes to build like a 10 solution, you know, a minimum viable product, I think that's a rare skill. A lot of people are stuck kind of in their niche. Um, and when you can combine those two, it's like, okay, here's someone who can build prototypes that have, you know, something visual—could be internal, doesn't have to be perfect—but some kind of visual component and some interesting kind of back-end component, and then like motivate that prototype. 0:29:04.365 --> 0:30:01.859 So that a team can work with them to take that prototype into something that creates value. You know, those when you start like finding the intersection of these very rare skills, that's when you start to get someone who is truly mythic. Um, but yeah, it's—I'm totally with you. I think, you know, the idea that like you have this person who like—you know, can't is non-verbal but like sits in your basement and writes 10x as much code as other people, like that never really made sense to me either. Um, so it's good that people are continuing to evolve that. Um, all right. So okay, Patrick, I want you to watch this video. I'm gonna—I'm gonna read the title while Patrick watches it and gives us his live reaction. It's called 'The Beauty and Challenges of AI-Generated Artistic Gymnastics.' Oh dear. 0:30:03.344 --> 0:30:08.676 Oh, oh! This is Will Smith eat spaghetti. That's right! 0:30:09.503 --> 0:30:12.692 It's—it's that, but it's with 0:30:12.692 --> 0:30:29.129 The—oh, which makes it even worse. It's like better than that. It's like version two for sure. Like it's much better than that, but it's so much okay. I just watched a gymnastic shaped thing fall through the floor. Okay, so what? 0:30:31.002 --> 0:30:36.080 I don't. This is nightmare fuel. My friend, why have you done this to me? 0:30:39.068 --> 0:30:56.340 So like at some point it's like, wait, but it's clearly obvious what it is. Like you can see legs and arms and gymnastics leotards—I don't know what you call them—but none of like if you pause any one of these frames and just showed it to someone, they would not think that it would just be bad AI. 0:30:56.340 --> 0:31:27.027 Yeah, the difference I think what makes this so much next level beyond the Will Smith video is the quality. It's so good, you know? In a sense of like, you know when you do see somebody's face, it looks good. It's just the interplay there. There's just so much—I don't know how to describe. Everyone needs to take a moment, go to our show notes, click the link, and watch this video. It's only a minute, but it's how does. 0:31:27.685 --> 0:31:36.983 This only has 31,000 views? Like did this get ripped off of like TikTok or something? Oh, here from Instagram. I see it looks like—oh no, I can't tell. Oh yeah, you know. 0:31:37.000 --> 0:32:03.080 I found this in an AI think a machine learning Reddit or something. I'm not sure exactly where I found it, but yeah, the let me see if we could describe this. Okay? So basically what this is is an AI for folks who aren't gonna watch a video or listening in in their car or something. It's a—it's an AI-generated video of gymnastics. So imagine, you know, the parallel bars, uh the what is it? 0:32:03.080 --> 0:32:49.440 Called when you, the balance? So more specifically, if you ever watch it like during the Olympics, it's like that exact setup. So the stadium, right? You know all like the there's like judges on the side. So it's like a very conventional gymnastics meet or competition or whatever you call it. Yeah, and then there's like gymnast on a balance beam, and you'll see a frame where the person is standing like at the start of their routine. Then they begin to do their routine except instead of bending down and doing a handstand with their legs flipping up in the air, their head folds into their stomach and they go into a gymnastic shaped blob that then falls through the floor and separates into two pieces. And but the fans are in the background like they're still bleachers. The scene has coherence. They look billboarded or something. It's very weird, but the scene has coherence. It's just. 0:32:51.385 --> 0:33:00.970 Like it clearly—like yeah, it doesn't understand whatever it understands about humans, the gymnastics are breaking those assumptions, right? Right. And and you. 0:33:00.970 --> 0:33:36.306 Know there's this contrast between the gymnast who is like, you know, just morphing between all these like aberrations. I mean, there's one where it's just like two sets of legs connected to each other that are flipping over over each other. But there's this contrast between that and the people—you know, there's gymnasts who are preparing or there's judges, and those people because they have pretty low emotional frequency, they're pretty normal. But there's this person in the middle who's just completely you know like going through some crazy motions. Yeah, and. 0:33:36.627 --> 0:33:55.949 The equipment although it bends weird like tends to also be well rendered. Um but right, and you see people in the background yeah like oscillate. Like there's a judge in a suit, and then it transitions to a person in a white shirt sitting at a score table. Like there. Yeah, the more you watch this, the weirder. I gotta stop watching this dude. All right, I'm out. Yeah. 0:33:56.050 --> 0:34:04.994 I mean it's one of these things you can just every time you see something different. But yeah, I think it's a—okay, actually, I remember my point. Okay? So is it. 0:34:05.196 --> 0:34:08.824 Uncanny Valley related because that's what I'm getting the vibes of. 0:34:08.875 --> 0:35:22.703 Yeah. Oh, it's definitely got uncanny valley vibes. My point is, I'm seeing a critical flaw. I saw it in 2012 when we did the Alex Debt stuff, and then you know, I saw it again with self-driving cars also around 2012. And then I'm seeing it again where people who aren't in the AI field—especially investors, you know, a lot of people with a lot of influence who are generally you know really smart people who are counted on on their expertise, right? They fail to see that AI is this extremely multidisciplinary field. And being good in one area doesn't mean you're good in all of them. So if you have superhuman image detection, that doesn't mean you've reached superhuman AI. And I think probably what we'll see over time is is AI like be less of a focus and us to like break down AI into more of these subfields, right? Because with self-driving cars, you know the detecting of stop signs and pedestrians and these things became superhuman right early on. Um and. 0:35:24.390 --> 0:36:26.069 And so people thought, oh well, you know we're days away, right? But the reality is is the policy like knowing when you can break the rules, knowing like okay there's a set of road cones here, but that really means that that whole area is off limits. Like the road cones are kind of telling you this lane is shut down. Like all of the common sense there we're well behind humans, but because we're superhuman on the image stuff, people didn't—people kind of translated that. And so you know I think we talked about inpainting and how good it is and everything in the beginning, but it's good for people to know that like there's still like entire fields of AI where where um we're just massively massively behind. And I think this kind of shows that I think you know merging AI with kind of something that's like physically plausible or just having continuity is something that we don't really know how to do very well yet. 0:36:28.684 --> 0:38:19.975 This one, yeah. I mean, I think there are lots of extensions, so gymnastics specifically probably is very difficult because the poses are very uncommon in the training data. Like, the people are to some extent—I mean, it's superhuman's wrong word, exceptional human. Like these people are very elite in a very specific set of skills, so they're doing things that are far outside, like standing on your hands as an example. If you looked at imagery across human, you know, civilization, the amount of people standing on their hand at any given time is just super, super low, right? Yeah, but during a gymnastics routine, you spend huge amounts of your time, like hands down, feet up, right? It's yeah, so there's problems with that. And then, like you said too, I think there are all these layers and layers of nuance and semantics that just—yeah, it I don't know if it's the same, but it feels similar to when you know my software is done and it's like, well maybe like you never we are very bad sometimes at identifying the quote-unquote hard part or the part that'll take the most time. And so you get people really excited, and then you yeah kind of see it tip over and fail. But the inpainting example is a good example. Like, you inpaint this way—we're saying—and if it fails, it's very low risk. You just do it again or you try in a different way, right? That's very different than I have some security camera and there's a tree in the way, and so I'm going to inpaint the tree and like use that as like anything could show up there. People are going to get confused. Like it's a very, you know, risky endeavor. Like that—that's not you would go, 'Oh, that obviously isn't going to work.' It's like, well maybe I mean, yeah, that case seems pretty obvious, but to some people, you may show them that and they may be thinking like, 'Oh, it can see through leaves,' and it's like of the tree, and it's like that's not. 0:38:19.975 --> 0:38:27.029 Really how it works? Yeah, I'm going to inpaint my stock portfolio and it sees, 'Oh, there's this Enron-sized hole,' but you 0:38:27.434 --> 0:38:39.803 Probably could find examples where it worked great for whatever reason, just by happenstance or whatever. Like, yeah, you inpaint and it predicts the stock trend, and then you think you can just inpaint as a way of doing stock 0:38:40.343 --> 0:38:55.180 trading. Yeah, you know, this is a hype cycle I've seen within like my kind of local community, and I'm seeing it now. And I feel like—I don't know if anyone from my workplace listens to the show. I mean, I work at a pretty small 0:38:57.252 --> 0:39:42.932 company. But there's even people I work with now who are really good at AI, brilliant, and they think they can predict the stock market, and they do well as long as everyone else is doing well, and then they get completely crushed. I mean, I don't want to wish any future ill will to ill gains to anybody that I know, but I've just seen this so many times now that when someone says, 'Yeah, you know, I'm also doing some algorithmic trading, you know, on the side,' and oh, it's like starting to make X amount of money, and it's like at some point I don't need to work because I'll just make X, and at some point X will become enough.' And then there's always some kind of market downturn, and almost every time people get hit really, really hard. Yeah. 0:39:45.160 --> 0:40:19.872 I pay a fair amount of attention. I mean, we talked about on this thing to finances and the stock market and that kind of stuff. To be convinced out of what you're saying, just go look at the kind of people who work at Renaissance Technologies and the kind of proven returns that they get, and then say, 'Can you outcompete those folks?' Um if you don't know we don't have time now, but like yeah, the kinds of people Renaissance Technologies hires for their hedge funds—and Jim Simons and like—yeah. There's just yeah, I don't want to go toe-to-toe with the kind of folks that work 0:40:21.357 --> 0:40:56.727 there. Yeah. I mean, just okay. So to play devil's advocate, you know, you could argue that there's also a lot of really uninformed people, but I think it's, you know, it's not the kind of situation where you just have to be better than the average trader. I think you have to be a lot better than that to have a guaranteed stable high yield. And so just just saying like, 'Oh, I know AI and I can use it and be better than like the 50th percentile trader.' Probably isn't enough to quit your job,' for example. Yeah. 0:40:58.617 --> 0:41:08.573 I would say it a little different, is but kind of the same same idea is if when you talk to these people and I'm no, I'm no expert, and I'm not trying to gate you and you do whatever you want. I don't really—I don't really care. 0:41:08.860 --> 0:41:11.745 Like yeah, it's not financial advice. If you lose all your money, it's 0:41:12.015 --> 0:42:27.025 not our fault. A lot of the success stories are, you know, survivorship bias. One that's true. They were doing this at a very different time, you know, 40, 50 years ago. I think like you said, yeah, there's a lot of the meme stocks, but if you see the people who go up against on the other side of GME—the GameStop—and how wrecked they get even though that they're well financed like trading against that, you know, maybe even now. But then the thing is I think a lot of the people I hear saying what you do go to it's like a certain corner of the internet, but experts don't hang out there. And so you get a lot of other people who are talking about algorithmic trading specifically—it's a really bad one and Python scripts and things you can share and bots, and people are more than happy to sell shovels to gold miners for technologies and data sources and this kind of stuff, and they're not illegitimate business. I mean, they're just trying to help people. But if you go spend time listening to, you know, industry podcasts or people who are sharing a little bit and just the amount of consideration—so not just volatility but the curve of volatility as you go through the expirations and options and how hedging and the various Greeks and the various inefficiencies or efficiencies of options versus futures, and when you start talking to most people who are doing the kind of like, 'Oh, I see,' you know, up into the right returns—they 0:42:28.628 --> 0:42:54.548 have to make an argument that that stuff is irrelevant, that it doesn't matter. All that matters is number go up, and that's a very reductionist view that could be true, but most likely isn't. Um and so you're playing a gambling game, and if you want to play that, that's fine, but you probably need to go learn about all those other things and at least say they don't apply or you've considered them and determined them to be irrelevant rather than I didn't even know. 0:42:56.590 --> 0:43:06.209 That was a thing. Yep, yeah, exactly. All right, we'll go to book of the show. Patrick, you're first. What's your book of the show? All 0:43:06.664 --> 0:45:05.262 Right, mine is The Three-Body Problem. By Oh, okay, we're gonna try it. No, no, I don't know. Okay, Zizhen Liu. The Oh, I okay, I can't say it. All right, you can look it up. The Three-Body Problem. Um, this was a book I had read a while ago. I thought it must have talked about, but I looked it up and I don't think I have. And it is written by a Chinese author, which is super interesting in science fiction because I feel like it feels very different to a lot of Western science fiction that's written. But it recently got made into a series by Netflix from the people who did Game of Thrones. But that makes it better or worse for you is your own sort of opinion. But The Three-Body Problem itself is, I think it's a trilogy. I only ever read the first book because I actually didn't love the book. But here this TV series actually makes it a little bit more palatable, I think than the book, which was pretty tough to read for me. But definitely an interesting take of, you know, during the Chinese Revolution, someone just basically like down on humanity. It's not good. Like there's lots of problems and ends up communicating with an alien race that, you know, doesn't share that with other people. And then when people find out there's this whole like how is human want to relate to the fact that there are aliens coming to Earth? And then it sort of plays out from there. I read the summary, so you may consider that a spoiler, but that was all stuff contained in the back of the book. So um yeah, interesting setup, interesting problem. There are all sorts of other like, you know, semi-nearby people talk about. Um, you know if if you kind of believe that Earth is on a catastrophic course due to global climate change in some way, could you really just release tons of fertilizer into the ocean to sequester carbon dioxide by causing a giant algae bloom? And it really is when you start to look. 0:45:05.819 --> 0:45:36.987 At it feasible that a very, very, very small group of people could take something that has absolutely catastrophic results for Earth or for humans kind of like just on their own that there doesn't need to be a, you know, group agreement or a large decision-making body. Um and it's it's kind of an interesting take. And so definitely recommend it. I just finished the Netflix series. I also thought they did a good job with it. So if you're looking for something to watch, it's definitely a pretty serious show. So um just know that going in. But yeah. Oh. 0:45:38.404 --> 0:45:47.382 Cool. Um yeah, I will actually check that out. So it's on Netflix? Is it also called The Three-Body Problem? Yes. Oh, very cool. Um um yeah. So. 0:45:50.149 --> 0:47:50.140 My my book of the show is is also a book that I wasn't crazy about. So I guess Patrick and I have something so just you know breaking the fourth wall here, listeners. You know, Patrick and I we read a lot. I don't know if we read more than one book every two weeks. I mean, that's pretty hard for at least for me to do. And so we're gonna have to review whatever we're reading. And so it sounds like this week we read some books we're not totally crazy about. I will say that. So my book is The Checklist Manifesto. Um I really enjoyed the stories. Um so this person is a doctor, a medical doctor, and so a lot of their stories were just really interesting medical situations. Um there is a lot of medical or maybe what's the right word? Like I mean they are like cutting people open and doing all sorts of stuff, and there is some detail there. So if you're very squeamish, obviously there's no pictures, right? But if you're squeamish, you know think about that before you get the book. But um but I thought it was fascinating. I mean, it's a there's a whole world around patients and complications that can happen and miraculous survival stories, resurrection stories, people who are completely incapacitated and doctors were able to revive them. Um The point of the book is that you should have checklists and force everyone to use them. And and you know it's like a pretty simple point. I mean, you know it definitely doesn't need a whole book. Um the book doesn't, you know, it kind of motivates through example, which is a little strange because it's kind of um you know if you give an example someone can't really argue that. Like if you're like, okay, I added a checklist and then we saved this person's life, the argument against that is what like. 0:47:50.140 --> 0:48:44.822 Maybe maybe you would have had some extra time on your hands. You could have eaten an extra meal or something like it. It's like it's like almost indefensible or unassailable. Um so you know I wouldn't look at it as a particularly powerful self-help book or anything like that. Um but these stories, I think were captivating. Um I got this one off the from the public library using the Libby app, which is I don't know if it's only an American thing, but um depending on where you live, you might have an app that connects to your local public library and lets you—it's effectively like a Kindle. Um a Kindle or an Audible app that's connected through the library, and all the books are you borrow them. Um and so I found this one on there. I thought it would be pretty interesting. Gave it a read. Overall kind of lukewarm, some good stories, but I wouldn't recommend it. 0:48:47.759 --> 0:49:59.984 All right, Tool of the Show. Well, to say something, I actually did enjoy. I just finished playing Not To 100, but I finished The Super Mario Bros. Wonder. Um I don't know that I've talked too much about Nintendo Switch games. Um so this is a little bit off topic, definitely not a tool, but I really enjoyed this game. To me it was just like for where I am in life, it's the right complexity. It was fun and challenging. I didn't just you know rip right through beat every level. Some levels were a bit of a challenge, but nothing was too challenging. So you know anytime I did sort of a session where I sat down and played the game, I was able to make good progress. Um and it is not super super long, so I was able to feel the accomplishment of beating it and the sort of style and approach and whimsy of those the sort of 2D platformer but rendered in a 3D view and the kind of goofiness. It doesn't take itself too seriously. It's not dark. It's just sort of very light-hearted. It was a very enjoyable game to play. Um and so if you've not checked it out and you have a Nintendo Switch—I mean, I'm sure you've heard about it—but I definitely give this one a thumbs up. I really enjoyed. 0:50:00.557 --> 0:50:12.977 Playing it cool. How does it compare to there's a Super Mario? I think it's called Super Mario Wii 3D or something, but it's on the Switch? Have you played Super Mario 3D World? Or oh yeah, something like that? 0:50:13.163 --> 0:50:48.381 One is 3D, I think if it's the one I think about. I think there's another Super Mario Bros. New Super Mario Bros. game as well or something which I didn't get into The 3D World. I really liked but it's in 3D, so there's a bit of a—I don't know—there's a different thing to playing 2D. It sort of allows them to be more creative in the limitation. And so I kind of maybe I just old school. I kind of prefer the 2D over 3D and I don't like when the games get like super hard. There are certain Mario games, I think, get very punishing in their complexity. Yeah. 0:50:48.820 --> 0:51:10.808 One I was thinking was actually called Super Mario U Deluxe? Okay, that's what was like Mario and Luigi. Yeah, yeah, that's the one I have now. It's it's okay. Um I think it is pretty difficult. I mean, I was able to get through it, but my kids are get pretty frustrated with it. Um I wonder if Super Mario Wonder is a little easier? Yeah. 0:51:11.230 --> 0:51:25.810 I tried that one. I didn't get into it nearly the same for whatever reason. The clicking there's probably different development teams. I know I'm not into it enough to sort of look, but this one was definitely very well done. Oh cool. Maybe I'll. 0:51:26.367 --> 0:53:21.572 Check it out. Um all right, my tool of this show is Amazon Q. I feel like maybe I think we had it as a news last time. That's right, we had a news and I've been using it pretty heavily. And so I'm gonna upgrade it from a news to a tool. But yeah, this uh this uh it needs like it just ate a mushroom and now it's super super news. Um I I'm recently building some stuff. I I I'm not ready to release it yet, but I'm getting there. So I have to keep it all under wraps for now. But um um but maybe another thing I I used Amazon Q for was the LinkedIn bot. Um I don't know if we talked about the LinkedIn bot? Yes, we did. Okay. So I use it for LinkedIn bot. I'm using it now on this new thing I'm building, and um it's amazing. Like you know some things it does really well are um if I'm refactoring something, so if I'm taking some code out of a function—yeah, taking some code out and making it its own function, then I'll copy cut that code, you know, create a function, paste it there. And then I'll go back to the place where I cut the code out, and Amazon Q knows like call the function. So it's like it's like it just like auto-completed that for me. Um and so there's there's just like yeah, there's so many things where um and when it hallucinates, you know, the way it works is it's just like auto-complete where you get this shadow of what it wants to write, and if you press Tab, you know that shadow becomes a reality. And if you do anything else, it just it just disappears. So you kind of get used to this like phantom code in front of your cursor. Um and over. 0:53:22.129 --> 0:54:14.661 Time you get used to sort of being blind to it when it's doing something that you don't really expect, but then you get this like wonderful surprise when it actually does auto-generate exactly what you wanted. And you'll even, you know, sometimes it'll all generate whole paragraphs of of, you know, like maybe five, six lines of code. Um so yeah, I think it's amazing. Folks should check it out. Obviously as we talked about in the past, you know don't do this for work if you're doing this on a work computer. You should instead go to your boss and ask him to get some kind of license or figure something else out. I'm sure there's companies that have enterprise solution there, but but if you're if you're in college, high school building something on your own, it definitely behooves you to get an AWS Builder account and get Amazon Q. Um the 0:54:16.602 --> 0:54:51.077 The reason I ended up on Amazon Q and not GitHub Copilot or one of these other ones is frankly just better marketing. You know, I have the AWS extension a lot of people have that in their VS Code, and they just parlayed that into a, you know, like a little notification. Like I think originally was part of the AWS extension. So you kind of everyone just got it for free, and then it became its own extension once they had kind of a critical mass. So they did a great job marketing it. I think that all of them would probably be equally good. 0:54:52.866 --> 0:55:06.417 Though yeah, I'm excited that when these become commonplace and able to be used at work and there's like a well-established practice around them. But yeah, for now, I've not delved into these, but I definitely kind of want. 0:55:07.851 --> 0:55:17.480 To. Yeah, I mean, I'd be curious for your coding for C++ and these things if it would be as useful. My guess is it'd probably struggle. 0:55:17.480 --> 0:55:21.880 Oh no, no. Don't tell me that. I want to envision it being a panacea. 0:55:24.405 --> 0:55:40.032 Yeah, I mean, well okay. So a couple of things that are just first principles, right? Like it's not going to know like oh this memory address is actually like a FIFO in hardware or something, right? So oh, okay. Well, yeah. 0:55:40.032 --> 0:55:59.300 Fair enough. But oh, I mean that would be the day when you can feed an AI—I don't have to do that quite as much anymore, thankfully. But if you feed it a data sheet and it would build out for you communications protocol like that, that would be amazing. Oh man, that'd be awesome. That'd be cool. All right, now I'm just fantasizing. All right, somebody out there make it happen. 0:55:59.300 --> 0:57:02.200 All right, on to the topic of resume writing. This came from a gentleman. I never give last names because I don't know who wants that or not, but Matthew C. Um we had a discussion on LinkedIn. It actually Matthew sent me a LinkedIn message. He said that you really enjoyed the show. He was wondering if I could take a look at his resume. Fortunately, my LinkedIn AI didn't flag it as spam, which it has flagged—actually, it has a couple of false positives. So if you're listening, I do apologize for that. But this one, you know, let through. And so um and so yeah, I talked to Matthew. I gave him some tips on his resume, and I said, 'Hey, what if we did a show on this and gave you credit?' And he was thankfully, you know, really supportive of us doing that. So thank you, Matthew, for inspiring the show. 0:57:04.373 --> 0:57:11.005 To start at the maybe obvious: why even have a resume? I mean in them in. 0:57:12.979 --> 0:59:12.623 Back in the good old days, no. They were that good. It is actually a little—you'd show up and you know you may have gotten a job interview in some various ways, and then they would ask you for a printed out, and you would have like a little leather or at least we—that's what we always have, like these little leather binders, and you would unzip them and pull out, you know, on a printed very nice paper that you would buy specifically for printing resumes, and you would, you know, hand them over. I haven't done that in a very, very, very, very long time now. Most of the time everybody gets copies electronically because in order to apply for a job, you have to go to some portal on some website, and the almost very first thing you need to do is upload your resume and then proceed to type in most of the information from your resume into various forms. But you know, it's an expectation that you have a resume when you apply for jobs today, and the reason they have—you know, upload it is to share it with everyone, um to have it recorded for other potential openings. It is a thing, but it's also because they do run keyword detection. They try to categorize it. And when you kind of read on the internet, you see people bemoaning this, and I do agree it's—it's kind of problematic, but it's also very difficult, especially if you work at a large company and get many thousands of resumes trying to categorize so that the right people look at the right resumes. Because as a person who does hire sometimes, you'll get a dump of resumes, and if they're—if they're mismatched, it becomes very difficult to trudge through them and say like, 'Hey, these are totally the wrong kinds of people.' Like it's not that they're good or bad; it's just they're not this is not a match for the kind of role that I'm looking for. And so I need to find a way to get access to the ones that I need. And so I think having a resume, probably pretty obvious, but I'll give one non-obvious thing, which is as you. 0:59:13.534 --> 0:59:46.643 And maybe you're already there. I don't want to make an assumption, but one of the things is as you move through your career, it becomes a little harder to remember some of the exact details that you want to provide for previous positions and roles. And so it's really a log. You may trim stuff off as they become older and older, but having those things written down so that next time you go to apply for a job, you're sort of appending and then trimming down is actually really, really useful. To have that sort of log that you can go through and be able to curate for future applications. 0:59:47.892 --> 0:59:49.545 Yep, yep, yeah. Totally. 0:59:51.098 --> 1:00:25.000 So what is a resume? There's actually not that much variance. Most people follow, you know, pretty standard formats. Sometimes I've interviewed designers and more creative folks who have also done interesting things to their resume. But in a lot of those cases, those folks also have a portfolio. So you really can't go wrong on your resume. But for them, but for professionals, I think you know try to keep it to one page. 1:00:26.856 --> 1:00:39.344 I think that's important. I think you know that way everything is just, you know, fitting on one screen. And it usually starts out with your name and some generic information. Some people have put, you know, whether 1:00:40.845 --> 1:02:32.320 they need a visa or not? You know, that can be an important point that sometimes I've seen on the very top line of the resume. It'll usually just some basic information: where does this person live, their name, etc., phone. And then it goes straight into various education and work experiences and accomplishments there. And then usually at the end it will end with a list of let's say skills or trades that this person has. So this person might be an expert on compiler optimization, and so they'll put 'compiler optimization,' you know. And sometimes people will put more soft skills and things like, you know, led their volleyball team in college or something like that. This is a little different than a CV. So a CV, which I think is Latin—Curriculum Vitae, something like that, I think that's right—a CV will list out, typically listing all of your research papers, or depending on how many you have, it might be a subset of them. And so that's where it can take several pages to list out. But even if you're writing a CV, I would effectively have a resume on the first page and then have a section of notable papers and awards, and then you can have a list, and that list might be four, five, six pages, but it's basically a glossary of your favorite research journal papers and awards. So you're still basically writing a resume and adding some content to it. 1:02:32.320 --> 1:03:20.027 Yeah, I would agree. If for some reason you need to go past one page, and some people say it's okay to be on two, sometimes people have a very strong preference for one. And if you're going to two, unless there's a really good reason, it tends to just draw on too long. So if you have stuff that requires more, I would say prepare two different—I think that's what Jason's alluding to as well—as like having two different documents: one is your resume, concise, one page, and then you can even say on there, 'You know, more details or CV available if needed,' or just something. And then have another document which is longer and goes into more details if there are really things that you need to list out. So here's another one though: Jason, I don't know how you feel about it. Some people say having a GitHub on the resume—a GitHub link—is a requirement. 1:03:21.141 --> 1:03:30.580 A good question. I think okay, I think you should only put things on your resume if they're good. 1:03:30.777 --> 1:03:31.722 Is a great. 1:03:31.924 --> 1:03:57.574 Answer. Okay, yeah. I mean, I think this is—your resume is not your autobiography, right? You're you're maybe your autobiography is also rose-tinted, but if I was to write an autobiography, I would definitely call out all the extremely dumb things I've done in my life along with all the things I'm proud of. So this is not that. This is supposed to be the absolute best side of you—no flaws, nothing. And so... 1:03:59.295 --> 1:05:18.878 And so your GitHub could be the best part of your resume. And I think if it is, you could have an entire section dedicated to that. I've seen that work to really good effect where they'll say, 'Here's a Personal Projects section,' and the header will have their GitHub, and then they'll have a bunch of projects. And if they're popular, they'll have stars. If they're not popular, a good thing to do is to say kind of what technical problem this solves and maybe what projects it's similar to. So if you built, for example, an automatic differentiation engine and an DSL language for that, then you could say, 'You know, similar to PyTorch,' and so you know you built your own, obviously. It's hard to eat into the market share of PyTorch or TensorFlow, but you built something on that in that milieu. So yeah, I think generally it's really important to have that kind of portfolio, just like any artist or creative person would. But don't just put your GitHub and you have like zero repositories or something—that's not helpful. Yeah. 1:05:20.194 --> 1:06:23.360 I agree. I've seen, and I know people who say you have to have your GitHub, but there's a lot of personal preference here. But if you put your GitHub, I think it's you're saying there's important to have spent space on your resume to put. So I'll go visit it, and if I go visit it and it's like, 'Oh, you just have various forks of projects that require you to fork it for some reason,' or like configuration files, or just things that are obviously class projects where you started with the template—yeah, it just doesn't feel like what this isn't helping me. It's not really hurting me, but I would just leave it off and save the person the time if there's nothing there that you're really trying to show off. But I think it can depend. I think some people don't hesitate if you have stuff even if it's not having a lot of stars or isn't really popular, if it's code that you wrote that you worked on, you should be proud of it and you should put it. But if it's code that you just copied or forked or is a group project, I don't know. 1:06:25.720 --> 1:06:48.640 Yep, yeah, I totally agree. I mean, the thing too is if you don't put it, then people won't think about it. It's a 'strisand effect,' right? So if you put the GitHub repo and, yeah, as you said, it's all like University of Chicago Course 12, you know, zero stars and there's no tests or anything, then now people like, it's actually kind of worse than just not putting anything. 1:06:48.640 --> 1:08:33.859 All right. Well, I think we're maybe going through some do's and don'ts. So I'll start off with a couple of do's. Definitely be specific. So when you're writing that, 'Hey, I was at this company,' I mean, that's very specific. But the dates that you were there, the location—those things are good. You can put the skills used. I think that's important too. But also like when you write projects, you know, say specifically things that will make sense to people who may not be super familiar with the jargon of what you were doing. So if you were doing something in a database but your database had a particular internal name—probably not super useful if you're applying externally—to put the internal names of those databases. So put down, 'You know, database schema design,' right? But definitely be specific about not the work that was being done on the project, but the work that you were doing on the project, right? And so we see I see a lot of people that you spend time asking because they're pretty junior and they put a pretty high level project, and it's, 'What did you do? What were you responsible for?' And I think—and it's not again, you don't have to be nervous or ashamed or bashful about it—but just put down the specific thing that you were doing. 'I was writing unit tests.' 'I was doing integration,' whatever it is that's fine. Put that there. Don't put, 'You know, I was responsible for a project that did $100 million and was 10,000 lines of code that had an SLA of 99.999 uptime.' It's like, okay, this sounds like a group of people were doing this. What were you doing? 1:08:33.859 --> 1:09:49.840 Yeah, it's a good call out too because it'll give the person looking at the resume some idea of your kind of compass, right? So for example, if someone says, 'You know, I implemented the Python Standards Committee at my company and I made like 20 or 30 decisions around different syntax and all these things,' then it's going to—excuse me—it's going to paint a picture to interviewers that you're somebody who really cares about DevInfra and the developer ecosystem and all of that. You know, conversely, if you say, 'Hey, you know, I wrote this code that added this much incremental revenue to the bottom line of the company and I worked on these features that caused this many more customers to come in the door,' then people know, okay, this is somebody's very product focus, and they're going to be a good person to be in between the business and the engineering.' Right? So isolating your specific contributions helps the interviewer know what your strength is. Yeah. I mean, 1:09:52.253 --> 1:10:50.286 I think—I guess I don't know kind of correlarily the sort of thought there if you—I mean, I'll be a little harsh, but if you try to scam your way into getting the interview, people are going to figure it out, especially in the kinds of jobs that we're talking about here. Almost always there's going to be a technical portion of the interview, and it's going to become obvious. So even if you happen to put something down to score an interview—either when they go to check your references or they're talking to you—it'll come out. And so you'll just have wasted time, and then people are going to be irritated and upset. Or you're interviewing for a role that's higher level than you'll be able to successfully interview for, and sure you may tell yourself, 'Well, there's like a very outside chance that I'll score it.' I don't know. The opposite is worse, right? Like the opposite that you get sort of kicked out of the queue because people are like, 'No, this person is just—there's just no good,' rather than, 'Oh, they were miscategorized.' Right? 1:10:51.062 --> 1:11:49.800 And even like at a more granular level, you might just not get the right job for you. Like maybe you really should be spending half your time with these Python committees, you know, setting the course of the company, like and the language specs and all of that. But you know, you like really tried to fluff up like your impact numbers, and because of that now you're on some product team you could have been on some other team—it was a better fit, etc. So it's like, you know, represent yourself. I guess it's I want to make sure that we're not creating contradiction here. Like obviously, you know the time you accidentally deleted the production database, you know don't put that on your resume. But you know of all the good aspects of you, you know be faithful to that sort of distribution of talent and represent that as faithfully as possible on 1:11:51.559 --> 1:13:04.020 The resume. Another thing is any item that you put on your resume, be ready to talk about. So if you put down, hey, you know I've seen Goof—I'll give a goofy one first. Someone put down oh, I can solve a Rubik's Cube in under a minute. And then so the interviewer thought it'd be funny they brought a Rubik's Cube and asked the person to solve it. And so I mean that's a bit of a goofy, goofy example. And they just put it in there—be funny—but they got called out on it. But I've seen people put down, you know, I'm an you know I'm an expert or you know I know and then they put down six programming languages, and then someone starts talking to them about one of them like, 'Well, I used it for a tiny project once.' And it's like, well maybe you needed some way to qualify that or call it out. But similarly, it's very frustrating when someone puts down that they worked on a, you know, project that's relevant to the role that they're interviewing for at a company and you ask them to talk about that project, and I've had people just told me, 'I don't remember,' or 'I don't know.' And it's like, why did you put it on your resume? Like, you have other stuff here. Like what is it that you do remember, right? Like be ready. If you put it on there, people are gonna ask you about it. Yeah, that's 1:13:05.120 --> 1:13:12.306 A really, really good point. Um yeah, don't put that, you know, 20 languages unless you've been listening to the show for at least a 1:13:12.306 --> 1:13:13.133 Few months. There we 1:13:14.989 --> 1:13:23.157 Go. Um how many languages do you put down if you were going to put them down? Jason, I'm curious like what what threshold would you have to cross? Like how many do you think you'd put? 1:13:24.237 --> 1:14:27.434 You know, oh, that's a great question. So okay, I have maybe a slightly different view than Patrick, although it's very similar. So I think that you can put down any language as long as you're prepared to really understand it between now and when you're interviewed. So for example, let's say you're interviewing for a game development job, right? So and let's say you have like a you've done a few small projects in C# right? So you can put C#, but if your resume gets through the screener and you land the interview, you're going to need to take a week off whatever you're doing and cram C# really, really hard because like you don't want to show up—I think it's okay to flex a little bit on the interview as long as you're prepared to back it up by either knowing that language really well or getting like really up to speed with it very quickly. Okay? 1:14:28.125 --> 1:14:56.965 So I will accept this that you say that if you're willing to at least do one of the technical interview programming problems in any language you put down, right? Yeah, we're on the same for the interview. Okay. I will buy this premise. All right. So how many—oh, how many for me personally? Yeah, what would you how many would you put again? Not saying like you might be plus one because you are trying to target a specific job and it's close enough to something you've done. I'll buy that, but like how many would you feel comfortable? Yeah. So here's—here's what I think for 1:14:56.965 --> 1:15:18.059 Me personally, obviously different for different people, but I think that if I cover like the three or four most popular ones for my profession, so my case that would be Python, C++, JavaScript/TypeScript, and what would the fourth one be for AI? Um I would 1:15:20.438 --> 1:15:52.028 Actually, you know, I would probably put something to the effect of like Bash, Unix kind of scripting. I mean, I haven't thought about this in a while, something like that. All right. Um and so I would put those and then I would just assume that companies know like okay all of our AI is in Swift, but we know that like yes there's a lot of people who are really smart who don't have used Swift yet, and so they're going to have to interview Python people. That's that's the way I would approach it. 1:15:52.669 --> 1:15:56.500 Yep. If they wanted it for AI, they would—they would just let you learn Swift on the job. 1:15:57.563 --> 1:16:22.251 Yeah, exactly. Yep. Um now if you're yeah go ahead. No no, go ahead. I was gonna say if now if I'm like let's say I'm junior engineer, you know straight out of college, I would probably try to figure out what language the company is that I'm applying to, like what language they need, and I would put that language and I would study it hard if I got the interview. That's what I would—that's what a 19-year-old, 22-year-old, whatever Jason would do. So all right. Fair? 1:16:22.977 --> 1:16:24.540 Enough. So I would put two. 1:16:26.419 --> 1:17:06.683 And I hear your points. I'm not saying no on them, but my reason for bringing this up is that you said three, four, okay? I give it the Bash thing, okay? Maybe everyone could add that. I don't know; I don't actually consider my Bash scripting to be that great. So maybe I would have two tiers and in total have maybe four—maybe maybe three. But we routinely see people out of college with six, seven, eight programming languages written down, and I'm just like, no, there's no way. It's like, you know what is the bar here? If I asked you even just basic questions in a random one of these, what is the chance you're actually going to be able to answer? 1:17:07.476 --> 1:18:00.632 These. Yeah, it's tough because for example, if you're applying to—you know, I work at a self-driving car company. If you're applying to do work on the car, you're going to need to know C++. And so literally people who don't have C++ on the resume, they just the resume just doesn't make it through. So if C++ is like you've done a few examples—you know, you've done—I guess I think the part where we agree is you need to be able to show up and do a 60-minute interview in every single language you put on there. Now, I think that there it's probably not too hard to get to the point where you could do a technical interview in six languages. It doesn't mean you've had years of experience in any of them, but yeah. 1:18:02.809 --> 1:19:01.517 I guess yeah, this is an interesting topic. There's some variance here. If you in your example—I mean, I've hired people who I needed to know C++. I'm going to ask them, 'Oh, okay, use this hash map, standard unordered map. What's the default insertion policy for standard unordered map?' Hey, what is that? Like, you know, what is this thing going to look like? What is if you have this kind of error that equates to a symptom of a stack overflow? What is your suspected reason why? Or I'm going to write code up on the board that introduces an array index out of bounds and ask you, 'Hey, what's going to happen here?' And the answer is undefined. Then we should talk about what are potential things that could happen.' So oh, see, that's totally different. See what I had in my simply no—that's an extension that's a yeah. If I needed a specific language and I wanted to check that you actually know that language, but yeah. So this one, this one's interesting. People should be thoughtful about. 1:19:02.209 --> 1:19:09.360 I mean in my mind, I had this mental model where like you're using one of these tools to interview, right? Like— 1:19:09.360 --> 1:19:14.160 You're saying like yeah, I could write LeetCode problems in one of these six languages. Yeah. 1:19:14.160 --> 1:20:14.046 Right? But you're totally right. If well, so that's an interesting thing, right? Let's dive into this. So if someone puts—let's say Python as one of their languages, but they're a junior engineer, you know, they graduated last year from college, right? Or they got a bunch of Coursera certificates last year. Or let's just stick with C++. They let's say they only have one language and it's C++. They'll probably still not be able to answer those questions, right? So I think it's—it matters whether someone says they can do a line. So there's basically like three levels: there's 'I bet I can learn this in a week,' to where I could be as productive as anybody else, like any other person who is not an expert. Then there's like 'I could fix weird problems.' And then there's like 'I'm a guru.' It's kind of like there's three levels, probably more than that. 1:20:14.046 --> 1:20:16.290 Don't put guru on any of them. Just— 1:20:17.202 --> 1:20:37.553 As this, don't ever put your guru. But you're right. I think if someone sort of transcends the number of languages, right? If someone just puts C++, they're really not telling you whether they can pass the kind of questions that you just asked. And I wonder if there's maybe a way that in their resume they can. 1:20:38.633 --> 1:21:46.820 Right? So I'll amend—I'll amend what I'm saying based on what you did. I think when you list them out, maybe having tiers, like, you know, I don't know, not as 'I'm sure it's like proficient in,' and then maybe a couple languages, 'you know, proficient.' I'm trying to think of like wordsmith it, but like 'proficient' and like, 'you know, familiar with,' and then maybe a few other languages. Oh yeah, right, something like that. But I'll amend—I think you're right. I wouldn't be harsh. So I'm not going to bring someone in and ask these kinds of questions unless we didn't talk about this when you're listing out your experiences. You should definitely write what languages each one of them was primarily in, or which language or languages each of them was in. So I'm going to look if I need someone with actual years—not out of college, not, but someone who says they are a C++ programmer because I go look in their resume and it says they spent four years in this project doing C++. Then I'm going to expect them to have some day-to-day—I mean, I was trying to give extreme examples to evoke a response. But like I'm going to expect them to have some familiarity if they've listed that. That's that. 1:21:46.820 --> 1:22:13.909 Totally makes sense. I think that's actually the way to do it because all these things like expert, guru, proficient—they're all very qualitative, right? But if you were to say C++ for 10 years straight, well then yeah, I think it's fair to ask that person like which of the different standard maps is better in which cases, and whatever. Like that must have come up in the past 10 years. So yeah, I think that's a good way to go about. 1:22:13.909 --> 1:22:16.680 It. We've made progress. The world is a better place. 1:22:16.680 --> 1:23:35.584 That's right. So don't on your resume—the biggest one is don't have mistakes. I had a not a mistake, but something that was confusing on my resume where and I think you probably had this too, Patrick, where you know I was a student and I was working at the same time. Now for Master's degrees is very common, but it's not very common for PhD students to also work. And so this really threw people for a loop. And you know, what ended up like really helping here was that I put very specific dates. So originally I put, 'You know, I started my PhD in—in 2004 and I finished it in 2010,' or whatever it was. But instead, I put really specific, like, okay, this is a day I started. This is a day I got a Master's degree. This is a day that I worked, and then this is the day I finished my dissertation.' And so like I think just being really specific—like ambiguity is a form of a mistake when it comes to your resume. And being really specific helps a lot. And definitely don't have mistakes if you put just from personal experience. If I see someone and they like the dates. 1:23:37.160 --> 1:25:37.101 Don't really make sense. Like they kind of work two jobs at the same time and they didn't work, and the dates aren't really lining up, especially if they don't match their use of experience. Like 10 years of experience, but you only worked for three years. I'll just say this isn't like—this is actually kind of a segue. I'll go into briefly. Managers have to look at a ton of resumes, right? And typically we'll only need to hire one, two, maybe three people at the most at any given time. And so the goal isn't necessarily to spend an unbelievable amount of time to get the absolute best person and squeeze out every single percent of the interview experience to be absolutely flawless, right? And this is knowing this actually helped me a lot even as an interviewee. Like I'll get rejected; I don't apply that much to jobs anymore, to be honest. But there was a time where I was really—you know when I was changing jobs like a few years ago. I was applying a lot of jobs, and in the past, sometimes I'll get rejected for really weird reasons or things seem like they'll be going fine and then radio silence. And what I've realized is that the system is trying to trade off, you know, the manager's time with having the perfect possible interview experience. So if you have a bunch of mistakes on your resume, the manager will say, 'Like this is going to eat up a lot of my time. I can just skip this resume and get somebody who's probably as good and save myself from having to ask all these clarifying questions.' Right? So mistakes are really not good. Try to avoid them. Look at the resume, make sure that you don't have a cover letter that's for the wrong company. I've seen things like that, and just be really diligent. It's important. 1:25:39.413 --> 1:26:45.500 I think the other thing that is a mistake is—I have seen grammar mistakes are not, and I'm okay with that because not everyone has English as their first language. But sometimes like spelling mistakes to where you said like it creates ambiguity. It's like I don't actually know what they're saying here. And especially if you're looking for someone on it, like they did this one thing and this one thing tells me maybe they have prior experience here, and it's written in a way where I can't parse what it's supposed to be saying. Like Jason said the company I do right now—like there isn't even really a good mechanism for me because I can't reach out to that person directly. So I have to go back to HR, ask HR to reach back out and like, 'Hey, can you...' Like this is not a thing that likely happens unless I really thought there was good potential in a resume.' And so you can get kind of bumped out for relatively small. So have people who aren't like just family have your family read it and just say, 'Like is this reasonably clear to you even though you may not understand some of the terms? Do you see any spelling mistakes?' You know that kind of thing. 1:26:45.500 --> 1:27:21.085 Yep. Another very simple one: don't put images. My company—yours probably too, Patrick—pulls the text out of the resume, and images really mess that up. I've seen this where it's just the information is just totally corrupt, and then you have to click on the resume and all that. And so there's probably folks who didn't even make it through the screen because of all sorts of weird artifacts that came out of the images in their resume. So just avoid that. There's not really a reason to. 1:27:21.810 --> 1:28:50.927 Do that? Related is—and we're going to do this in a second, but I'll just combine them here. It's not just images, but people will use like colored sections, like oh, you know gray and then a darker gray for like each experience of work to get visual separation. Looks great, I don't know. Whatever some of the—I think they get printed on a fax machine and then scanned in, but sometimes we end up with them that are super hard to read. The contrast is destroyed. The text extraction could fail under such circumstances. If you go to a job fair and you're handing a physical resume, sometimes like that's the one that goes in the system even if you upload one later. So just be aware. Like in my opinion, two things: one, don't use a fancy template with lots of image bullets, and think like just as simple as it can. It needs to look nice, but it doesn't need to be complex to be looking nice. And try not to—don't certainly not using any colors besides black and white would be my opinion. Try to keep it black and white. Try to make sure it's like the most parsable, like the easiest to scan. And if someone gets a black and white copy or a you know decimated version of what you uploaded, so really, really do keep it simple. And this is specific to our slice of—of, you know, engineering. If you're an artist or applying for, you know, a summer job, some—you know, I don't know. Those probably have their own sort of ways. People like, you know, maybe a spritz of cologne or something on their resume helps? I don't know. But none of that's going to help here. And like it's just going to. 1:28:51.670 --> 1:30:50.655 Create more problems. Yeah, totally. I mean my college experience—I built a lot of different things, and I really wanted to show that off, but you should send people to landing pages for those things. So my resume was all black and white, but it had hyperlinks to the pages which were then had some nice like design on them. But yeah, I think the resume is just not the place for that. I just think that you have to appeal to both the people who are looking at it and this sort of lossy partially automated autonomous system that's going to do all the pre-screening. And so you have to be careful there. Another thing I wouldn't do are, you know, put these like qualities of your character. This is actually—I mean, it would be really nice to show your character on your resume, but the problem is it's just, you know, when people are reading resumes and they come across like 'hard working' or 'friendly' or something, at best they just ignore it. Like they have blindness to it. And this is our 30th resume they've looked at today, and it's just—they skip over it. At worst, it does seem like a little bit weird, right? Because it's just it's just weird for someone to call themselves 'friendly' or 'motivated' or 'hard working.' I think again similar to what we said with languages, you know, if you're hard working, then find objective things that you did. So for example, let's say you built a hundred houses a year for Habitat for Humanity—like physically built the houses. Well, then that shows you're super hard working, right? So just put that. So you know, find like how have these social qualities resulted in some objective thing that you've done in the world and just put that thing. 1:30:52.107 --> 1:31:40.875 Yep, I agree. And I mean this extends to the interview. We're not talking about interview prep right now, but everybody's on their best behavior on resumes and on the thing for the most part. So if anything, so two things: one, everybody's saying they're hard working and motivated and willing to work long hours or whatever. They think like—I'm not saying anyone should or does do that, but like whatever they think will you know get them the job. Everyone is pretty much trying to do that. And on the other side, the people doing the interview assume those things. So they assume that people are on their best behavior and are that way. And so like Jason said it's really better to show than to tell. And if you need to put it, like don't make it too flowery. Like just because then people think you're kind of like overcompensating for something. Like why are they saying they're so hard working? You know, it's just it's very weird. Yep. 1:31:42.259 --> 1:32:00.079 Yep, totally agree. I think—you know, we didn't really have this in the do section kind of, but we'll cover it now is I do think it's important to call out your what do you call it? What would you call this section? Basically like I think it's like hobbies and activities. Yeah. 1:32:00.079 --> 1:32:01.395 Your extracurriculars. 1:32:01.395 --> 1:32:27.535 Yeah, extracurricular. Yeah, exactly. So you know one thing I put on my early resumes was that you know I ran like a volleyball team. I played in like three different volleyball leagues, and so I gave people the impression like, 'Okay, here's somebody who can be on a team, can lead a team, can like play a sport pretty regularly for a long period of time.' And I think that way I didn't have to put things. 1:32:29.290 --> 1:32:32.024 Like motivated or something or friendly or. 1:32:34.352 --> 1:33:14.447 Something. So yeah, definitely I think extracurriculars are good. I wouldn't put things that are controversial because again it's like if something is really polemic, right? And half the people will love you, half the people will not like you. I don't think the people who love you are going to pass you in the interview if you're not competent, but I do think the people who don't like you will make your interview really difficult. And so it's kind of a lose-lose situation. So I would avoid putting things that you know are just like polemic like that, but for things like sports, you know, charitable things—I mean these things are great for your. 1:33:18.244 --> 1:34:05.140 Resume. So I think a little bit about briefly resources how to produce your resume. So because of all the other things we've already been saying, I think it's pretty much any reasonable text editor would be fine, but even just using Google Docs to produce it. It really shouldn't have requirements that it be fancy and needs to be done in Adobe Illustrator or something. I mean if you want to do that, it's probably because you're looking for a job related to that. But for our field at least, I think it should be able to—you know Google Docs is free on the Mac, Asia's pages, or Microsoft Word, just something that in general is text focused and does allow you a little bit of formatting. I mean do format it to look nice, but don't go overboard. 1:34:05.815 --> 1:34:59.000 Yep, and just a general rule if you're doing something, you really want to think about the audience, right? And so this is a I'm going to link a podcast that I listen to that's geared towards managers. It's literally called Manager Tools: A Podcast, and um there they have an episode. It's actually a five-part series on how to scan resumes, and so like putting yourselves in the shoes of someone who's looking at your resume, I think can really help you in writing your resume. So if they say something like, 'You know, if there's too many colors, just throw it in the garbage,' and you're looking at your resume that you created in Adobe Illustrator—yeah, it'll kind of put things in perspective. Um so, yeah, I think this is a great chance to see kind of what other people see when they look at your resume by listening to that episode. 1:35:00.473 --> 1:35:17.365 That's great. And then I think you put this on here, but I mean for the programmers out there using LaTeX to do their markdown version of the resume, I guess is an option. It's going to get made into a PDF anyway, so I guess that's an option. Is that how you do? 1:35:18.057 --> 1:35:51.580 It. Yeah, you know, the oh, the reason I used LaTeX is for a CV. It helps with like the references that now I remember. Um but I wouldn't recommend that even if you're a PhD student. There's probably better ways to do it nowadays. I think the important thing is what Patrick said: you know, use the tool you're comfortable with and the tool that is pretty popular, and don't try and do it in an image processing tool and create something that's really ad hoc layout-wise. All right? 1:35:51.580 --> 1:37:50.440 Well, the last statement here, I think we're running a bit long, but hopefully this is useful. I think for a lot of these things, one of like Jason was mentioning is if you want to say you're hard working, show your hard working. Put down things that indicate that you're hard working. And I think it isn't the only goal, but when you are picking projects to work on at your job or careers or hobbies to do, thinking about how they may show on a resume is worth considering, right? I think not because the resume matters a ton, but because you're thinking about that as part of career development and part of having a career is, you know, having hobbies outside of that. But having the ability to show your potential as an employee. And so I think putting thought into how am I going to choose what to work on and what is it going to look like later? So, as an example, if I was reading something this morning—oh, Cobalt, there's a you know dearth of Cobalt engineers. So at work someone might say, 'Hey, we have this Cobalt code; we really want you to take on.' You should think: is that something that I really want for my future? Is it something that and a way of thinking about that is that something I really want to put on my resume? Like am I going to want another job in Cobalt? I mean, to be fair, it could be really lucrative. It's an interesting bet if there really is going to be a scarcity of Cobalt engineers, but if it's not something that's going to make sense there, then you might want to find and see if there's an alternative or something else within reason, not again as maybe the only thing you think about. But trying to consider how that might look. People do this all the time for university applications. They think in high school: what should I be doing to get into a good university? You're going to work a job probably a lot longer. Can work a lot longer. People can work four, five, six, seven years, so even putting a little bit of thought towards over that time, how do I compound things that will help me be marketable? 1:37:50.440 --> 1:38:18.045 is very important. And I think as a manager it's something that I try to think about for my employees is I'm not trying to retain them by making them unmarketable. I'm trying to make them feel like they can stay here by continuing to let them do things in a way that would be very marketable because then they're going to feel no need to—that they're trapped, right? They're doing things that they know they could parlay into other careers or other jobs other places, which makes them content and happy and growing. Yeah. 1:38:19.581 --> 1:38:20.965 That makes a ton of sense. 1:38:20.965 --> 1:38:21.927 Yeah, I mean, 1:38:23.631 --> 1:38:58.039 That's a great call out. I think just the last sort of call to action is you're if you're in college, you might think that you're the work to kind of like build yourself up is kind of over, like oh, I get my job and I never have to look at this resume again. I can just throw it in the garbage. But no. Like you're I think it's always good to kind of keep that in mind. I mean, definitely there's several different reasons to take on a project or a job, but that's definitely one that you should keep up there. 1:39:02.106 --> 1:40:24.300 Cool. All right. This is awesome. We covered a lot of good material. If folks out there have other resources, other tips, feel free to throw it on the Discord or on the Patreon group. Thanks everyone who have given us their patronage on Patreon. We really appreciate it. I did ask in the members-only section what kind of perks do folks want? Do folks want to have like a Zoom call with us? Do they want us to look at their resumes? Like what sort of Patreon benefits are folks interested in? And you know, I think there's as Patrick used the word earlier, sort of a survivorship bias here where the folks who are in our Patreon really just want to thank us for the content that we give everyone for free. We do really appreciate that. But I'll ask to the general audience as well if there is something interesting that you think we can do as part of the Patreon experience. Always just throw us an email, throwgrowingthrowdown@gmail.com, or you know throw it on the Discord or in Patreon. Thanks a lot. Thank you everyone. 1:40:47.300 --> 1:40:53.580 And share a liking card.