The Weekend AI Engineer: Hassan El Mghari

00:00:00.000 | -

00:00:02.000 | - In this talk, I'm gonna walk you through

00:00:16.520 | some of my projects that I've built

00:00:18.520 | and all of the lessons that I learned along the way

00:00:21.480 | to build great AI apps that can scale to millions of users.

00:00:25.200 | So, let's get right into it.

00:00:27.120 | So, to set the stage with some context,

00:00:29.120 | I've been building side projects pretty consistently

00:00:31.760 | for about two years now.

00:00:33.400 | And so, last year, I built about 11 side projects,

00:00:36.520 | and they got about 20,000 visitors total.

00:00:39.200 | So, not too shabby.

00:00:40.860 | So, my goal for this year was to try to double that number

00:00:43.200 | and get to 40,000 visitors.

00:00:45.740 | And happy to announce that I did hit that goal

00:00:48.500 | and slightly exceeded it as well.

00:00:51.000 | And-- thank you.

00:00:53.520 | And basically, here today to talk about how this happened,

00:00:57.440 | and, you know, I'm very thankful and very lucky

00:01:00.440 | that I managed to hit such a good number,

00:01:03.920 | over 8 million unique visitors across all of my projects,

00:01:07.280 | 20,000 GitHub stars and about 2.8 million people that signed up.

00:01:11.520 | And fun fact, every single one of these projects that I launched

00:01:14.720 | was built on the weekend.

00:01:16.720 | So, I'm going to pick through some of these projects,

00:01:19.120 | and we're going to go through them

00:01:20.360 | and talk about some lessons learned.

00:01:23.200 | I also want to mention that everything I do is open source.

00:01:26.560 | So, you can check out all of my projects at github.com/nutloop.

00:01:30.560 | Embarrassing gamer username from, like, 10 years ago

00:01:33.000 | that I can't get rid of.

00:01:34.000 | But, yeah, no, I love building in open source,

00:01:37.000 | and it makes me so happy to see people use my projects.

00:01:40.480 | But it's also a very good growth lever when you launch.

00:01:43.440 | And I get a lot of genuinely helpful PRs from a lot of people

00:01:48.120 | that are better at prompt engineering than I am.

00:01:51.120 | So, it's always helpful.

00:01:53.280 | Disclaimer, I do have a bit of an audience on Twitter,

00:01:56.360 | which is very helpful, but honestly, I don't think it's as important

00:01:59.520 | as people make it out to be.

00:02:01.360 | A lot of people can kind of attribute having a lot of followers

00:02:06.920 | to having successful projects,

00:02:08.000 | but I've seen plenty of people have very successful side projects

00:02:11.760 | with little to no Twitter following.

00:02:13.640 | And, in fact, less than 5% of the traffic of those 8.5 million people

00:02:18.040 | that have visited all of my projects,

00:02:19.640 | less than 5% of that traffic actually comes from my Twitter account.

00:02:23.200 | So, you may be thinking, where does this traffic come from?

00:02:25.640 | And, honestly, it's a lot of word of mouth and Google and SEO,

00:02:29.120 | and other influencers sharing it.

00:02:31.120 | So, I'm going to get to that in a bit as well.

00:02:33.800 | So, today, I want to talk to you all like friends,

00:02:36.920 | and when I talk to my friends about my projects,

00:02:38.720 | I kind of just share my laptop and go through a bunch of things.

00:02:43.200 | So, I'm going to switch over to my laptop here

00:02:45.760 | and go through a bunch of my side projects.

00:02:49.080 | So, let's do that. Wonderful.

00:02:53.080 | So, this is kind of my first AI project,

00:02:55.720 | how I got into AI last December.

00:02:57.840 | And, really, it stemmed from this problem that we had

00:03:01.240 | where we had just run a conference last year,

00:03:03.080 | and we had several hundred photos out there in an image gallery.

00:03:07.320 | And, right before we published it, my CEO came up to me and was like,

00:03:11.160 | "Hey, we probably need to add alt tags for a lot of these images."

00:03:13.720 | And that would have been a very painful process,

00:03:16.120 | going through several hundred images.

00:03:17.520 | So, I looked stuff up, and I found a nice image-to-text API

00:03:21.440 | that ended up working really well.

00:03:23.240 | You know, I went and I checked a lot of these --

00:03:26.440 | a lot of the alt tags and maybe fixed, like, two of them and published.

00:03:29.080 | But, this is really my big, like, light bulb moment of like,

00:03:32.520 | "Oh, my God. AI can really, really help you save a ton of time."

00:03:37.680 | Like, this isn't some Web3 hype from last year.

00:03:40.320 | You know, this is real.

00:03:42.520 | No, I'm kidding. Web3 has its place for sure.

00:03:44.600 | But, this is really the big thing when it came out.

00:03:47.040 | So, I built this little open-source project.

00:03:49.240 | I put it out there.

00:03:50.240 | And then, I just started having fun and building other stuff.

00:03:52.240 | So, I built another project called QRGBT with my friend Kevin at a hackathon.

00:03:56.520 | And so, the idea is that you just generate just pretty nice QR codes.

00:04:00.120 | So, we can actually go and generate a QR code for AI.engineer.

00:04:06.200 | I forgot the domain name.

00:04:07.560 | And we can select a prompt here.

00:04:09.520 | I'm going to just click one of the pre-generated ones,

00:04:11.640 | a forest overlooking a mountain.

00:04:13.560 | And hopefully, in like five or six seconds, it should generate a QR code that links to the conference

00:04:17.560 | that just looks a little bit better than the black and white QR codes.

00:04:21.360 | And so, we built this.

00:04:22.160 | And we weren't expecting way too much.

00:04:25.640 | Because people really don't have to generate QR codes every single minute.

00:04:29.200 | So, yeah.

00:04:30.520 | We put it out there.

00:04:31.320 | We got about 8,000 visitors, about 8,000 QR codes generated.

00:04:34.320 | And so, we were like, okay, cool.

00:04:36.760 | And I was like, all right.

00:04:37.760 | I want to try to build something that has more, like, daily active users

00:04:41.280 | or people that will use it consistently.

00:04:43.160 | So, I built this little tool that summarizes TechCrunch articles.

00:04:47.320 | So, the idea is that you go to TechCrunch.com.

00:04:49.640 | You can click any article that you want.

00:04:51.760 | And all you have to do is add summary to the end of the URL over there.

00:04:55.000 | And it will redirect you to my website and kind of summarize the whole article using GPT 3.5

00:05:01.040 | in a couple of bullet points.

00:05:02.400 | And so, the reason I'm showing you a video here and not a live demo is because I got a very

00:05:07.040 | nice email from the TechCrunch lawyers when I launched this telling me to take it down.

00:05:12.640 | And so, that was a lot of fun.

00:05:15.120 | But, yeah, anyway, I took it down and I moved on.

00:05:18.480 | That one -- it did pretty good when I launched it.

00:05:20.120 | And then they made me take it down.

00:05:21.240 | And it kind of died off from there.

00:05:24.120 | And then I started just, like, replying to random people on Twitter.

00:05:27.120 | So, Samina here asked, like, can someone help me build an AI to help me take classes?

00:05:31.960 | And I was like, all right, bet, I got you.

00:05:34.160 | And I built this little thing in, like, a couple hours where it takes some information

00:05:38.360 | about yourself, your face shape, and your gender, and you can add some relevant context.

00:05:42.800 | And it uses a combination of LLMs and the Amazon API to find the ideal glasses for you

00:05:47.960 | and actually links them on there so that you can buy them.

00:05:50.880 | So, yeah, I just started replying to a bunch of tweets.

00:05:52.880 | Another one was by my friend Theo who said someone should make an app that kind of auto-generates

00:05:58.280 | commit messages for you.

00:05:59.680 | And then my CTO tagged me and was like, CeCe, I love that idea, which translates to build

00:06:05.820 | this as soon as possible.

00:06:06.860 | So, I was like, all right, I got you.

00:06:09.520 | And I built a little tool.

00:06:12.880 | So, essentially, you could run git add, you run the CLI tool that I built, AI commit, and

00:06:17.440 | it analyzes your git diff and produces a little commit message for you that you can then use

00:06:21.700 | to commit.

00:06:22.700 | And these are, like, very small, hacky solutions.

00:06:26.000 | You know, my CTO tagged me at 7:53 p.m. on February 11th, and then less than two hours

00:06:32.140 | later, I replied with that little script.

00:06:34.680 | Thank you.

00:06:38.860 | And after I saw it get some attention, I was like, okay, I need to clean this up.

00:06:42.860 | I need to figure out how to bundle it into an NPM package.

00:06:46.020 | And so that's what I spent my Monday morning on.

00:06:48.960 | I hope my manager isn't watching.

00:06:50.820 | But that was a fun Monday.

00:06:53.500 | And, yeah, I kind of bundled it out there and posted it as an NPM package.

00:06:56.960 | And now I think over 30,000 developers are now using it to commit their messages.

00:07:04.000 | And it's one of my more popular open source repos.

00:07:06.440 | There's some PRs that I need to take a look at.

00:07:08.880 | But, yeah, a bunch of 6,000 stars and about 25 contributors.

00:07:12.840 | And so this was kind of my exploration with LLMs.

00:07:16.700 | And so actually I have one more project called the Twitter bio generator.

00:07:20.140 | And essentially also open source like most of my other projects.

00:07:23.600 | But you just put in some context about you so we can do like engineer at Microsoft.

00:07:29.380 | And we can say loves volleyball.

00:07:31.340 | And pick a vibe and it will make your Twitter bio for you.

00:07:34.840 | And kind of stream in text from GPT 3.5.

00:07:39.600 | Spiking code bugs and volleyball balls.

00:07:42.500 | You can't get any better than that.

00:07:44.740 | But you might take a look at some of these projects and think like this is so simple.

00:07:48.840 | Like nobody is going to use this.

00:07:50.080 | This is just like this little chat GPT wrapper.

00:07:52.620 | Like everybody in this room can build this thing.

00:07:56.200 | But I think we constantly underestimate like the majority of the world are not AI engineers.

00:08:02.680 | Nobody can build this.

00:08:03.680 | A lot of people haven't even used chat GPT yet.

00:08:05.380 | Like it's crazy.

00:08:06.380 | So even the simplest apps can do really, really well.

00:08:09.400 | And so that's a common theme that you might see is like all of these are very simple apps.

00:08:13.720 | So I launched it and I got about 200,000 visitors that used it.

00:08:17.900 | I got about 100,000 people in a single weekend.

00:08:20.040 | And then I hit my open AI bill and I had to shut it down for a little bit.

00:08:25.200 | So it's always a good sign.

00:08:27.620 | And so after this I kind of switched into image to image model.

00:08:30.340 | So I built this photo restore website that basically unblurs old photos.

00:08:35.060 | And the motivation behind this actually was my parents sending me these old photos.

00:08:38.860 | So I'm actually going to put in a picture of my dad doing karate when he was like 18.

00:08:42.500 | And he sent me this photo and his face is really blurry.

00:08:46.020 | And you'll see.

00:08:48.220 | Yeah, he's flexible.

00:08:49.220 | I do not.

00:08:50.220 | I did not inherit that.

00:08:51.420 | But you see his face is a little bit blurry.

00:08:54.200 | You can't see it too well.

00:08:55.200 | But hopefully in the space of a few seconds we should see.

00:08:58.200 | And so this uses just a GAN model.

00:09:01.040 | It's called GFPGAN.

00:09:02.040 | It sends it to that model and it will basically scan like all the faces in a picture and restore

00:09:10.440 | it.

00:09:11.440 | So we'll see if the internet is working out.

00:09:13.920 | We'll hopefully see the image come in in a few seconds.

00:09:16.920 | And if not, I can move on and come back to it.

00:09:21.680 | All right.

00:09:22.680 | I'll come back to it.

00:09:23.680 | So, again, open source repo.

00:09:26.880 | And this one, like, really, really did well.

00:09:29.840 | And it kind of is my most consistent project.

00:09:31.760 | It still has about 250,000 people that use it every month.

00:09:36.700 | Mostly actually in India and Indonesia, which makes a lot of sense because the phone cameras

00:09:40.860 | there are a lot lower quality, so it makes sense that they would use it.

00:09:44.360 | But shortly after it went viral, I got a lot of inappropriate images being uploaded.

00:09:48.700 | And so I had to -- I used actually TensorFlow.js and I published this as a library as well.

00:09:55.400 | But yeah, I just ended up using this to scan the image and make sure it was safe before I

00:09:58.960 | processed it.

00:09:59.960 | So let's go back.

00:10:00.960 | Okay.

00:10:01.960 | So it looks like it was restored.

00:10:02.960 | We'll actually put them side by side and zoom in a little bit.

00:10:03.960 | So you can see his face before a little bit blurry.

00:10:06.280 | And then after the transformation, you can see it really, really clears up.

00:10:08.960 | Thank you.

00:10:09.960 | And really, another thing I want to stress here is that this is one single API call to

00:10:19.640 | this GFPGAN model.

00:10:20.640 | And that's it.

00:10:21.640 | And he's really getting that and displaying it back to the user.

00:10:24.780 | So it's such an exciting time to be an AI engineer and to build this stuff because it's so easy

00:10:29.960 | and it's so impressive to other people as well.

00:10:32.860 | So I'm going to talk about one more project, and then I'm going to start to talk about some

00:10:36.760 | takeaways.

00:10:37.760 | And before that, actually, this is like the architecture for most of my apps.

00:10:40.760 | Really, I use Next.js on the front end and the back end.

00:10:43.760 | And you saw for restore photos, there's this little upload component that I use.

00:10:46.860 | And so the user uploads an image.

00:10:49.440 | It gets sent to Cloud Storage.

00:10:51.000 | And then I send that image URL to my Next.js API route.

00:10:54.540 | Or you can think of it as just like a Lambda function.

00:10:57.080 | And then that sends it to my machine learning model, to GFPGAN, to get restored.

00:11:02.080 | It gets back the image, sends it back to the client, and display it to the user.

00:11:06.180 | So this is kind of the architecture I use for a lot of my image-to-image side projects.

00:11:10.180 | But my last one, which -- I'll restart.

00:11:14.280 | But my last one that did the best is actually called Room GPT.

00:11:18.440 | And it's that idea of if you give it a room -- I'm just going to give it a random living room

00:11:23.180 | on the Internet, and we're going to select a couple themes.

00:11:25.880 | But if you give it a room and some themes, the idea is that it will use this, and it will

00:11:33.200 | help you redesign your room.

00:11:34.460 | It will give you different variations of that specific room, different color themes, different

00:11:38.800 | like couch styles and stuff like that.

00:11:40.920 | So we can see it just finished.

00:11:42.280 | You can see it really respects the structure of the room.

00:11:44.880 | So it looks the same, but it gives you different ideas for these tables and backgrounds and tiles

00:11:49.140 | and everything like that.

00:11:51.060 | So really the motivation behind this project was that I saw somebody else built this before,

00:11:55.980 | but they used Stable Diffusion.

00:11:57.420 | And Stable Diffusion actually does a notoriously bad job at maintaining the original structure

00:12:02.780 | of a room.

00:12:04.040 | Like you can give it a room, you can tell it, okay, redesign this in this theme, and the image

00:12:08.640 | it produces looks nothing like the original room.

00:12:11.240 | Like the dimensions are messed up, the depth is messed up.

00:12:14.220 | And then I saw this new model called ControlNet that came out, and ControlNet does really well

00:12:19.340 | at maintaining that structure of the room.

00:12:21.200 | So I saw that, and I was like, oh, this could be cool to build.

00:12:24.260 | So I put it out there, and I launched it on Twitter, and obviously it's also open source.

00:12:30.620 | But I launched it on Twitter, and it did pretty well on there, and kind of kept tweeting about

00:12:37.120 | it.

00:12:37.120 | Because the thing about Twitter, when you tweet about something 24 hours later, it's kind

00:12:40.540 | of dead.

00:12:41.540 | So what I like to do is I like to kind of post updates over and over again.

00:12:45.540 | So we had about 10,000 people that used it in the first 12 hours, and then 30,000 in the

00:12:52.040 | first day.

00:12:53.040 | And then I added some testimonials, may or may not have paid these people.

00:12:57.940 | And then, yeah, two days later, I had like 90,000 people, and then three days, 270,000 people.

00:13:03.400 | And so it kind of just blew up.

00:13:05.260 | And I feel like it was just -- it was mostly because I was one of the first people to kind

00:13:09.160 | of productionize this ControlNet model that had just come out.

00:13:13.260 | So a lot of people were seeing it for the first time and using it.

00:13:17.040 | And most of these users, again, I can show you the analytics chart.

00:13:19.820 | So I have about 6 million people that have visited the site, and about a little over 2

00:13:23.140 | million that have registered and used it.

00:13:25.100 | And you can see the vast majority of the traffic is just Google.

00:13:28.580 | It's just straight up from Google.

00:13:30.540 | You know, a lot of people kept sharing it.

00:13:32.740 | And, you know, part of that, I think, is because it was open source, and a lot of developers

00:13:36.980 | liked it and re-shared it, but also the fact that I kept it free.

00:13:40.440 | So I'm going to talk about how I did that kind of when I transitioned back to slides.

00:13:46.240 | And so those are some of my side projects.

00:13:47.940 | One thing I want to call out is it's a really good idea to use AI-enhancing tools when building

00:13:53.140 | a lot of this stuff.

00:13:54.140 | So use GPT-4 for your code.

00:13:55.940 | We have an AI SDK that you can use over at Vercel.

00:13:59.060 | And we also have this product called V0 at Vercel.

00:14:02.620 | And so it helps you kind of generate UIs.

00:14:05.100 | And what's really cool is you can kind of see other people generating UIs.

00:14:08.000 | We can click on this one, for example, which looks like the Apple Notes UI.

00:14:13.500 | And we can actually fork it -- we can look at the code, which is cool.

00:14:16.200 | So I can copy all this code.

00:14:18.200 | But what's also cool is I can look at these templates or look at other people's code and

00:14:21.540 | I can fork it, similar to how I can fork a GitHub repo.

00:14:24.800 | So now this is mine.

00:14:25.800 | I can kind of add additional prompts to change it.

00:14:28.200 | Or I can click this button over here and actually select different elements within the page.

00:14:32.900 | So I can select this div and tell it, like, add three more notes and alternate their colors.

00:14:40.460 | And I can press enter and update.

00:14:41.900 | And what it will do is it will just re-render this specific div.

00:14:45.900 | It will stream in the data using our Vercel AI SDK.

00:14:48.600 | It will stream in these React components.

00:14:50.600 | And, yeah, hopefully it will keep going and add all this stuff in.

00:14:55.600 | And, again, as it streams in these components, it adds them inside of this code box over here.

00:15:01.600 | So I think it's still generating.

00:15:06.300 | But eventually, you know, it will add all of the notes here and we can go into the code and kind of copy and paste it.

00:15:13.300 | And we can also run a CLI command.

00:15:15.300 | You can see it scrolls down because it's still generating.

00:15:17.300 | Here's, yeah, note three, note four, note five.

00:15:18.300 | There you go.

00:15:19.300 | So added the five notes.

00:15:20.300 | I can go take all this code or run this command and get all the code and kind of iterate on UIs that way.

00:15:25.800 | So it's just a way to kind of prototype a very early UIs.

00:15:29.600 | So I'm going to go back to slides right now to talk about some takeaways.

00:15:35.300 | So use AI tools to move faster.

00:15:39.300 | I mentioned that.

00:15:40.300 | I mentioned the AI SDK.

00:15:41.300 | I mentioned V0.

00:15:42.300 | But there's a lot of really amazing libraries.

00:15:44.300 | I love using Replicate and Hugging Face and Modal and a lot of these other tools and Brev.

00:15:49.300 | There's a lot of really cool stuff you can use to kind of train your models or move faster when you're coding.

00:15:55.300 | So this is a bit of a spicy one.

00:15:57.300 | I always tell people, don't do any fine tuning and don't build your own models.

00:16:01.000 | And this is specifically for launching MVPs.

00:16:04.000 | Because, again, the purpose of this talk and everything is like building projects very quickly on weekends.

00:16:11.000 | So you don't have time to fine tune.

00:16:12.000 | You want to keep things very, very simple.

00:16:14.000 | If you can't describe your idea to me in five words, like, it might not do great.

00:16:19.000 | I have friends that come up to me that are like, oh, I want to build this platform for developers where they can connect them to clients and they can have their portfolios there and they can have a chat and they can have this.

00:16:28.700 | And I just, like, stare into them and I'm like, that's not going to happen.

00:16:32.200 | Like, that's not -- you can't build that in a weekend.

00:16:34.700 | You know, if you can't build a -- so what I tell them is just basically downscope to an MVP and then launch it.

00:16:40.400 | And even RoomGBT, when I launched that, I had so many machine learning engineers that DMed me on Twitter and were like, oh, my God, like, what models did you train?

00:16:47.400 | What parameters did you use?

00:16:48.400 | How did you get the data?

00:16:49.400 | How did you clean your data?

00:16:50.400 | I'm like, dude, I just used, like, an API off the shelf, you know?

00:16:54.100 | You don't need -- you can do so much with off-the-shelf APIs.

00:16:58.100 | Another one is use the latest models.

00:16:59.600 | I mentioned a big part of RoomGBT's success is using ControlNet, which had just come out a couple days before.

00:17:05.600 | Launching early and iterating is so, so important because you don't know what's going to do well.

00:17:10.600 | So if you can de-risk your projects, if you can get a project out in one or two weekends, and if it fails, so what?

00:17:17.600 | You can pivot, you can move on to a new idea, and you can just -- yeah, you can just move on to other things.

00:17:23.800 | And so -- and if it does well, then you can double down on it.

00:17:25.900 | Then you can add additional things to it.

00:17:27.800 | So I've found that to be great.

00:17:29.800 | Another one is making it free and open source.

00:17:31.800 | Making things open source is always great because people learn from it and are incentivized to share it and will open PRs to your project.

00:17:38.300 | And it will also get you a bunch of followers.

00:17:40.900 | You know, I gained, like, 25,000 Twitter followers this year just from posting a bunch of these open source projects.

00:17:45.600 | And they're just all developers wanting to learn and help me out.

00:17:49.100 | So open source is amazing.

00:17:50.800 | Making things free is a little bit hard, right?

00:17:53.700 | Because as we know, AI workloads are really, really expensive.

00:17:57.800 | And so there's a few ways you can do this.

00:18:00.000 | I kind of play to my strengths.

00:18:01.200 | You know, I have a Twitter audience, so I can go to companies and be like, hey, I want to launch this project.

00:18:04.800 | I think it will get X amount of users.

00:18:07.000 | Please give me some credits and I'll shout you out in the footer and I'll put you in the read me and all this stuff.

00:18:12.000 | But I've seen a lot of other people replicate this with no followers.

00:18:14.900 | And the key is to just build a very high-quality open source project, put it out there, put, like, a $50 limit on it.

00:18:21.000 | And when you run out, you can reach out to the company and say, hey, like, my project went viral on Twitter and it's featuring you.

00:18:26.200 | And the GitHub repo is open source, so when companies see this, they're kind of willing to give you some credit.

00:18:34.000 | So shout out to Replicate and Bytescale and Neon and a bunch of my other sponsors that help me keep a lot of my AI projects free.

00:18:41.000 | And the last lesson that I have for you is making sure your UI looks good.

00:18:45.000 | Nobody's going to use your product if it doesn't look good.

00:18:47.400 | That's just something that's been learned.

00:18:49.800 | And so I actually spend, like, 80% of my time on the UI.

00:18:54.400 | Even though these are, like, AI projects, most of the time it's on the UI because you need to make it look good.

00:19:00.000 | And if you're not a designer, you can just take inspiration from a bunch of different websites.

00:19:05.400 | And that's what I do.

00:19:07.200 | I'm not a designer, so I just look at, like, five other websites and I kind of steal a little bit of each site to make it look good.

00:19:13.000 | Because I don't know how to just come and make a website that looks good, but I know when something looks good when I see it.

00:19:19.600 | So that's kind of what I do.

00:19:21.400 | So very quick summary.

00:19:23.800 | If you do these five things, I think you can go very, very far.

00:19:27.200 | And lastly, like, I tell people to use whatever tech stack they want to use.

00:19:30.800 | I like the tech stack of, like, Next.js and TypeScript and Tailwind.

00:19:34.000 | It lets me move really quickly and then using Vercel for deploying my apps.

00:19:38.400 | Two final ideas, and then I'm going to get off the stage so better speakers can come and tell you about their projects.

00:19:43.600 | But I don't work 24/7, despite what you might think with all of that.

00:19:47.800 | I actually spend most of my weekends relaxing.

00:19:50.600 | But what I do is I work in sprints.

00:19:52.000 | So I'll take a single weekend and I'll just drop everything and go and try to put out a project.

00:19:57.600 | And then for the next, like, two or three weekends, I'll just binge Netflix shows and hang out with friends and live my life.

00:20:04.600 | So this has worked out for me, but when I say, like, I work all weekend, I mean, like, 12 hours Saturday, 12 hours Sunday kind of deal.

00:20:11.400 | You know, I kind of drop everything and do that.

00:20:13.400 | And so if you have flexibility in your life to do that, you can go ahead and try it.

00:20:17.400 | If you're married or have kids or have other responsibilities, you can experiment with what works for you.

00:20:22.400 | You know, you can spend a couple hours every weekend here and there.

00:20:25.200 | But that's what I do, basically, a weekend a month where I sit down and I put out a project and then relax for a little bit.

00:20:32.000 | So, yeah, moral of the story is I think, like, do what works for you.

00:20:35.400 | I'm just kind of sharing what's worked for me.

00:20:37.000 | And the final thought I want to put out there is that you need to, like, put in the hours.

00:20:42.460 | I think a lot of people DM me and are like, hey, like, I'm feeling really unmotivated because I'm trying to build these projects and they're taking me so much time.

00:20:50.660 | And, like, you know, how do you do it?

00:20:52.780 | Like, what's your secret?

00:20:53.500 | And the first thing I ask them is, like, oh, like, I'm sorry to hear that.

00:20:57.800 | How many projects have you built?

00:20:59.380 | And more often than not, they're like, oh, this is my second project.

00:21:03.900 | And I just stare at them and I'm like, you can't go to the gym for the second time ever and then look down and be like, where are my biceps?

00:21:12.240 | Like, where, where, it doesn't work like that, you know?

00:21:15.300 | You have to go to the gym consistently over months to see progress.

00:21:18.040 | And so the same thing happens with side projects and coding in general.

00:21:21.200 | And if you're an engineer, that's even better.

00:21:22.580 | I'm not an engineer, actually.

00:21:24.580 | I don't do, I don't write code for most of my time at work and I just learned to code a few years ago.

00:21:30.160 | So I think genuinely anybody can do it.

00:21:33.420 | You just have to kind of put in the hours and build and good things will happen.

00:21:39.000 | So thank you so much for having me.

00:21:40.760 | Thank you.

00:21:41.240 | Thank you.

The Weekend AI Engineer: Hassan El Mghari

Chapters