Auto-generated transcript So I imagine most of my audience is composed of junior developers and I'm sure most of you are reasonably concerned with how fast AI models are developing and what that means for your career. So I'm going to share some quick thoughts on the topic. First of all, if you keep doing the same things that you're currently doing, yes, you will get replaced. It is what it is. First of all, the reason senior developers are so much more productive than you guys, you juniors, when using AI-assisted coding is because they've already learned the best engineering practices. So think stuff like writing clean code, keeping it dry so not repeating yourself, handling edge cases, documenting and commenting on your work, breaking down large problems into smaller ones. These are all engineering practices that actually make you a lot better when using AI code. In other words, if you are good at all of these things then you can use AI to write much better code much much faster and that is why seniors are able to produce so much greater code at such a fast pace for instance documenting and commenting on your code breaking down larger problems into smaller ones these are fundamental skills of prompt engineering this is the kind of shit that a prompt engineering tutorial is going to talk about breaking down larger problems into smaller ones so that the AI model doesn't have context rot and it can actually solve those problems. These are engineering practices that actually make you better at using AI tools. And so if you have a senior developer who has spent years and years of his work life building applications and writing code, he's had a lot of time and opportunities to learn these core skills over here, right? Whereas if you look at a junior developer, they just don't have that experience, right? The way that a junior gets this kind of experience is actually spending those months and years building large applications in a real company in a production setting But the thing is since they juniors they don have those skills and since they also using ai coding tools they're not going to learn those skills because they're using ai to circumvent learning right they're using ai to bypass the learning and if you as a junior developer are using ai coding tools to skip the learning of these core skills over here then yes you are going to get replaced and there is nothing I can do to help you. So if you as a junior developer want to become irreplaceable by AI, at least the current state of AI, we don't know how much more powerful it's going to be three or six months down the line. But for now, if you want to become irreplaceable by AI, then you need to do the following. Number one, learn to be specific with your prompts. If you notice a senior developer giving instructions to his juniors or the people he's mentoring, you'll see that he's very specific with the tasks and the information that he gives to his juniors, right? And that is part of being specific with your prompts. It's just another one of those core senior skills that makes you even better with using AI models. So learn to be specific with your prompts. Number two, learn to handle edge cases. Don't just think about the happy best case scenario of how a feature is going to be implemented. Take the time to think about the edge cases as well. And you can use AI to help you with this. And the way you do that is you explain that feature to AI and ask it to come up with any edge cases or inconsistencies or any, you know, stuff like that, that you don't really think about all that much. You can use AI to help you with that, but make sure that you are actually handling the edge cases, right? And when you're using AI to write a feature or write some sort of code, make sure that you actually handle those edge cases there. Because a lot of times with AI, if you even give it a decent prompt, it's only going to implement the best case version of your feature it not going to handle edge cases and so make sure that these are being handled number three learn to write clean code if you don know what clean code looks like you not going to be able to teach the AI what clean code looks like It is as simple as that So take some time off to study how good clean code looks like and how it works and how you can write it and then teach AI to do that for you. You can write some sort of a skill.md file or something like that and you can let AI clean up your code for you. But fundamentally, just learn what clean code even looks like because if you don't know what it looks like you can't teach the ai and you can't write clean code number four learn to keep your code dry and do not repeat it this is part of writing clean code as well but it is especially important when you're using ai to write the code because ai is notoriously bad at this ai will repeat every single line of code that it that you have in your project it's going to reimplement all sorts of functions and functionality in your code base just because it can't find it already, right? And this happens more and more in larger code bases as opposed to smaller code bases. So you need to make sure that you keep your code dry. Make sure that when you're giving a prompt to AI, you tell it where certain functions or files are that it's going to need so that it doesn't waste time re-implementing them. A lot of people talk about how verbose and unnecessarily long AI generated code is and that is part of the problem over here. If you don't keep your code dry that's exactly what happens. So make sure you focus on that as well. Next learn to problem solve. Learn to take some complicated problems, large problems and then break them down into smaller problems. This is part of good prompt engineering practices as well as just basic problem solving skills right. You're not going to ask AI to build one large feature all in one go, right? AI is not going to be able to do that for you. So you need to be able to break it down into smaller and smaller problems And then you can go implement one part of the feature make sure it working make sure the code is clean make sure that the tests have been written And then you can move on to the next part of that feature and then the next part of that feature and so on and so forth until you complete the entire giant feature that you set out to build. So learn to problem solve, learn to break down complex problems into smaller and smaller ones until you and the AI can kind of just wrap your arms around it and get a sense of how to solve it, right? Lastly, learn to document your work and add good comments so that you can actually maintain that code over the long term. This is especially important with AI because, let's be honest, if AI is generating most of the code for you, you're not really reading it all that much. And if you are, you're still not developing the muscle memory that you develop when you actually write that code yourself, right? You probably know how this feels like, right? You build an application and you write all of the code manually by hand. You know, you type it all by hand. You likely know where everything is and how the code base is organized, where certain utility functions are, where the database handling logic is. You know all of it, right? But when you're using AI to write all of the code for you, you're not going to develop that kind of muscle memory. And so you need to make sure that you're documenting your work and adding some comments in the code as well but also just writing a lot of markdown files to ensure that you have a good understanding of your code base right the larger the code base gets the more documentation work you need to do to make sure that you can maintain this over the long term and the and the benefit of this is is that you'll have a bunch of markdown files that explain your code base and certain parts of it and then you can give those markdown files to ai and then it's going to write even better code for you. So I'm not going to waste any more of your time. These are the kinds of things that you need to learn as a junior developer if you want to become irreplaceable by AI. So start learning these and yeah.