How To Apply To 1000 Jobs In 1 Hour With Ai

Hey everyone, I hope you're having a fantastic day. In this video, I'm going to show you how to apply to a thousand jobs with AI in less than an hour. Let's get into it. So right off the bat, let me go a bit off script and just explain what the process of applying to jobs even is, right? It is extremely deterministic. You're just doing the exact same things over and over again. You go to a website or a job listing, you click the apply button, you fill out the application form, you upload your resume, and you click the submit button, right? there's some tiny difference in the steps every now and then but like 99% of the workflow is just this right like you pick a platform like I don't know LinkedIn or something go to the job section pick a job that you think you're a good fit for click the apply button upload your resume fill out all the questions and the forms and everything and hit submit so repetitive stuff like this right is very easy to automate with code and I'm going to show you how to do that in this video because I've done the exact same thing. And now here is a process you need to follow, right? Like I've drawn this really complicated, really well-made flow chart over here, and I'm going to go through it with you. The first step is to scrape the jobs, right? So you need some sort of a job source, obviously. You can do this with platforms like Indeed, LinkedIn, Remote.io. You can even do this with freelance platforms like Fiverr and Upwork. I've seen plenty of people have like scrapers that they run on Upwork, which just scrapes all of the jobs over there, right? You can do all of these things. But the problem is each platform has different HTML content and structures, right? So you're going to have to build a custom scraper for every single one of these platforms. That is a kind of a downside, but like it's just reality. You're just going to have to live with it, right? Now for actually building these scrapers, I recommend using a browser automation tool like playwright. This is what I use for all of my browser automation tasks. And it works really great for me. And to actually build this scraper, you're going to go over to any of these platforms like I'm on Indeed right now, you're going to go over here, you're going to create an account, and you'll get a big list of jobs over here that you can apply to, right? What you do is you take the entire source code of the page, the entire HTML structure, and you just feed that into your LLM or AI coding agent and ask it to write a playwright automation that can scrape all of the content on this page, right? You ask it to scrape all of the jobs because that's exactly what I did, right? I didn't write any of the code myself. I just asked the AI to do it for me, but I made sure to feed it the exact HTML structure of every single page that I'm trying to scrape, right? That is crucial because like I said, every single one of these websites has completely different HTML structures. You need to adapt your scrapers to every single one of them. And even though we're using AI to write all the code, if the HTML you give it is not accurate, it's going to write some really buggy, unreliable automations that can't scrape the data you want, right? So make sure you're feeding it the right HTML. Now, you don't necessarily have to go into, you know, the browser dev tools and copy this entire thing. This is just what I'm showing you as an example. What you can do is you can go over to your AI agent and just ask it to create a playwright instance that goes over to the Indeed website or the LinkedIn website and just takes the entire page HTML and just downloads it into a random HTML file on your PC. The AI can then analyze that file and just build the automation on its own, right? That's the thing. When it comes to building scrapers like these, you just want to make sure you're feeding the AI all the information it needs, right? And typically all it really needs is the entire HTML structure and maybe some screenshots of the page so it knows what it's looking for. And once you do that, you have a good scraper. It scrapes all of the jobs for you. That's all you really need. Now, two extremely important playwright automation tips that I want to go through before we go into the rest of the steps in this pipeline. These are going to serve you no matter what kind of browser automation task you're doing, right? Not just scraping but also filling out forms um doing any other doing any other task in your browser that you want to automate completely you're gonna like these tips are going to serve you very well in those scenarios the first tip is to always use a persistent logged in user account now what do i mean by that well platforms like indeed and linkedin and pretty much any platform out there require you to have a logged in user account before they're going to show you some content right and even for platforms that show you some content or let you do some actions without having an account having the account will actually make the experience a lot better and a lot more reliable because a lot of sites they have bot detection scripts and they gonna catch you using playwright and all these other things right but if you're using a logged in user account many of these websites will let you bypass the entire bot detection mechanism just because you have a logged in user account so they just assume that you're trustworthy right and they don't necessarily check your play right browser instance for bots right so use that and secondly always try to avoid css selectors because they can change very easily you know a website is updating their design they're tweaking some color or some layout or they're just adding new content or whatever they're changing the html a little bit css selectors can change very often and if you're using them your automation is going going to be very fragile, very finicky. It's going to break every now and then, and you're just going to be left babysitting the entire thing, right? So use role and accessibility-based selectors. These selectors are very rare to change, right? They usually just stay the same because technologies like screen readers and other tools for disabled people are using selectors like these, right? They're a lot more reliable than CSS selectors. So try to make sure that your automations use these selectors and avoid CSS and expat selectors as well. Now, once you've got a bunch of scrapers that are, you know, scraping jobs from all of these different platforms, the next thing you need to do is check if you're a good fit for those jobs, right? Because you've scraped a big bunch of jobs, how do you know if you're even qualified for them? Now, this is where we involve an LLM. And the way we do this basically is you feed it the job that you're applying to, as well as your resume and you just let the llm figure out if you're a good fit for that or not right like if i take you to my code base over here where i built this entire thing you can see i've got my resume over here in markdown because llms are just really good at understanding markdown as we all know and i've got this prompt over here this fit check prompt this is going to analyze the job that i'm applying for analyze my resume so it knows what kind of work i'm qualified to do and then it's going to give me a json output over here that's going to tell me basically how qualified i am for this job right whether i'm a good fit or not and the confidence score that tells me how likely it is that i'm a good fit and then just a one-line reason on why the llm made this judgment right so that's all you really need right you need the fit check prompt that tells you all these things you need the resume ideally in markdown format you could use pdfs or something else as well but llms are just you know they operate best when you give them text and markdown is the best form of it that you can possibly give it right and once you have both these things you just paste the job description job title etc etc over here in the prompt along with the resume you send it over to an llm it's going to do some complicated matrix multiplication and weird statistical math and produce some text output that you can use and that's going to be your you know fit check now Now, once you know if you're a good fit or not, like, let's just assume that you are a good fit and you can move on to the next step. Next, what you're going to do is you're going to tailor your resume, right? Now, quick caveat, you can add other steps over here as well, right? Nobody's stopping you. This is your custom app. You can add other steps over here as well if you want to, like you could, I don't know, pull the email address of the company and send a personalized email or CV or something, or you could do some other steps, like you could, I don't know, have an agent read the company's website and, you know, grab some information to help you personalize your application a bit. Like you can add any other steps you want over here, but the, like I'm just showing you a basic pipeline over here and how it's going to work, right? But you can add other steps and make this as complicated as you want to. Anyway, the next step is to tailor your resume, right? Now we're going to go over to my sub flow chart, my tiny flow chart over here. What you need to do is you need to grab your resume and the job again, and you need to send it over to an AI for processing. It's going to output another resume in Markdown file and you can use that as the, you know, new resume for this particular job. Now for this AI processing, you're going to need a good prompt, right? Obviously, you're not, you're going to want to use a really good prompt over here that actually teaches the AI how to write a good resume You can grab some tips and you know guides on this from the internet and and just paste them into your prompt right And this way the AI will actually give you a resume that actually like really good Because like if you give the AI a generic prompt and ask it to rewrite your resume it's not going to do very well, right? If you give it the entire job details and the job description and everything, it's going to do slightly better because it has the context of the job you're applying to. If you then also give it like a really detailed guide on how to actually write good resumes, it's going to do really, really well because it has all of the context it needs to get the job done, right? So yeah, once you have all of those things, you can convert that resume into, let's say, a PDF or a Google Doc or something like that, right? Because a lot of these job platforms, they want you to upload your resume in like a PDF file or a Doc X file, right? And this conversion process is pretty easy you know you can like there's packages in python and typescript that do the entire thing for you there's websites out there that can do it for you and you can automate that process as well very very easily so don't complicate this too much and if you don't know how this conversion process is going to work literally just go over to chat gpd ask it how to automatically convert a pdf or automatically convert a markdown file to a pdf or a document and it's going to tell you the exact steps and you can then use your coding agents to automate the entire thing for you, right? So all that being said, you've now got a tailored resume that's custom made for this particular job. The next thing you're going to do is go over to the job listing page, click the apply button and actually apply for the job, right? Then it's going to give you a bunch of forms. You're going to want to fill all of those forms using AI, especially when it comes to the custom job questions right now there's two different categories of uh questions that you could get over here let me just write them down somewhere uh you could have deterministic questions or you could have non-deterministic questions i'll put this in red just so it's a bit more easier to see the deterministic questions are just questions that have the exact same answer every single time right so questions like i don't know your desired salary right this is probably going to be roughly the same for every single job you're applying to right your location this is obviously going to be the exact same unless you're you know flying first class to a different country every single week um your years of experience working in this industry or with some specific tech or framework etc right it's stuff like these that's gonna be the exact same every single time right so don't waste AI tokens trying to you know answer questions like these that can just be hard-coded into the application code right however some of the questions can be very non deterministic right for example share some big challenge that you overcame or whatever a lot of job applications have questions like these right uh what was the most complex automation you built and just really you know long and complicated questions like these many job applications are going to have them and instead of answering these manually you can answer them with ai now again you're going to want to feed the ai the appropriate amount of context and and the way you do that or at least the way i've done it is you have like a document that explains all of your different experience and all of the different answers to do all of these different questions that you can have you feed that into the llm and you ask it to generate answers for these questions based on that document so i have a document in my code base called like common qa basically just a bunch of common questions that i found on most job applications and i've just written answers for these already and i put all of them in like a document that i can just reuse again and again right and it's right here so i have a big bunch of questions and really detailed answers to all of them over here in this document what i do is i feed the entire file into an llm i then feed all the questions of the job into the llm and i ask it to generate some answers for those questions based on all the content that i have in this file right and you're going to want to do the same thing now how do you create a document like this, I recommend just applying to like 10, 20 jobs manually, you'll encounter a big list of different questions that they asking you Some of them will go into a deterministic category right And for them you can just you know hard code the answers in the application code or have some like CSV or JSON you know file that has answers to all of those questions because they're deterministic, right? They're going to be the exact same every single application, right? So you don't need to waste LLM tokens. You don't need to waste time trying to ask an AI for answers to those questions. You can just have them hard coded in the application code. But for the non-deterministic questions that you encounter when you apply to these 10 or 20 jobs, put all of them inside a markdown file and answer all of them in really, you know, high detail in that markdown document. And then you have your common QA document. You can feed that into the AI, right? I'll just write it down over here. You feed that into the AI and you also feed the list of questions from the job application into the AI. And then you'll have answers to all of these different questions that each job is asking you. You're going to fill all of them up and you're going to hit submit and then you're done. You've applied to a job. Repeat that process with this automation a thousand times. It's just going to spam job applications that are hyper personalized because we tailor our resume that you're actually qualified for because we check that with an LLM and you can send a thousand high quality applications every single day. Like, yeah. Now, another quick caveat, applying to the jobs is going to be a bit finicky, depending on how you build your automation. Again, use Playwright and try to use role and accessibility based selectors. Where is it? Right here, right? These are just way more reliable than CSS selectors. So just use that and make sure you're logged in, right? Because most of these platforms have really convoluted, really messy HTML structures. And they have that on purpose to avoid having bots applying to their jobs, basically, right? So if you're trying to bypass that, you're going to want to have a really, really reliable browser automation tool and really reliable automation scripts that don't necessarily break every single time the HTML changes. So make sure you spend a lot of quality time on that. But other than that, yeah, this is all you really need if you want to apply to a thousand jobs a day. And again, all of these are high quality jobs that you're actually qualified for because we're personalizing our resume. We're asking AI to check our resume and we're, you know, asking it to go through all of our details and see that we're actually up and see if we're actually qualified for it or not. So yeah, do all of these things. You can apply to a thousand jobs a day with AI. And unlike most other AI, I don't know, educators, influencers, whatever you want to call them, automation specialists, unlike most other people that would recommend you this approach, mine is actually mostly deterministic. Like if you ask other, you know, AI influencers how to achieve the same thing, how to apply to a thousand jobs with AI, they're going to make you use an LLM for literally 90% of the workflow. They're going to make you use an LLM to scrape the jobs, to apply to the jobs, to fill out the forms, everything. You don't necessarily need all that. That is not only making your automation less reliable because you're asking a hallucinatory, non-deterministic LLM to produce the exact same output for the exact same prompt, which it's never going to do. It's always going to vary its output slightly. But even beyond all that, LLM tokens are very expensive. Code is extremely cheap. LLMs are not guaranteed to run the exact same way every single time. Code is guaranteed to run the exact same way every single time. So for the steps in this pipeline that can be done deterministically, we're using traditional code to automate them, which is just so much better than using an LLM, obviously, right? So yeah, this is the best way that I found to automate the process of applying the jobs. I hope this is helpful to you. And if you want to see more automations like these, definitely subscribe to the channel. And if you want me to share the entire code base that I've built over here. It's like, look, this shit works for me, right? Like I'm not just talking about this out of my ass. I actually built the entire thing. It works for me. I'm using it every day. And that's why I'm sharing it over here. So if you want to see the entire code, it's going to be in a GitHub repository. I'll link it in the description. You can go check that out. If you want to see more videos like this, or if you want me to go even more in depth in how I create automations like these that are extremely reliable and yet still automate really complex high reasoning tasks, subscribe and leave a comment down below on what you wanna see next. Anyway, that's all. I hope this was helpful to you and I'll see you in the next video.