How to Get Past AI Filters

[00:00 - 00:04] But now we're going to look at keywords. So this is the one that most people are probably aware of.

[00:05 - 00:22] And we talked about resume parsers, how they will look for specific keywords as selection criteria and potentially reject candidates based on that. So obviously identifying the core technologies and skills in the job description, it's usually written right there.

[00:23 - 00:36] So it's usually not super hard to find. Additionally, if you're applying for a role that you're very qualified for, you'll probably have an idea of the technological and competency landscape that will make you succeed in that role.

[00:37 - 00:48] So this is stuff like your coding languages, your Kubernetes, Docker, your tooling, JIRA, for admin stuff, other technologies like SQL, all of it, right? Some of these are make or break.

[00:49 - 01:00] Some of these are nice to have. So sometimes there are nice to have that are not evident from the job description, but which again, if you know the space, you might know it includes.

[01:01 - 01:17] If you've had a very long career, you probably have worked with a very wide array of tech stacks. You've probably worked with more things than you can fit in your CV while still making it easy to read and easy to parse.

[01:18 - 01:36] So I definitely advise being strategic about what you put into your CV. And one last thing, be very careful of spellings of keywords when you're talking about proprietary things and brand names, spell checks will often not pick up spelling issues.

[01:37 - 01:55] And you have some names that are maybe a little bit funny where you have the version of the software is important in some cases is not important. For example, it might be important, maybe less so now, but five years ago, it wasn't worth the specified Python three instead of just Python for some jobs.

[01:56 - 02:22] I think it's less important nowadays that three has become more dominant, but stuff like this is worth being an eye on. Another thing I've seen recently, I haven't been able to test it myself, but I saw an article about this where someone set up their own test jobs on LinkedIn, send a load of CVs through to see how that filtering was working.

[02:23 - 02:43] And based on this, it seems that having keywords woven into descriptions like into a sentence, as opposed to just listed in a skill section is more effective and tends to get picked up better by whatever LLM screening tools are embedded in LinkedIn. I can't speak for other platforms.

[02:44 - 03:02] Obviously, take this with a pinch of salt if I haven't had time to invest in it myself. But that does make sense based on how core summarization is as a function of modern LLMs and how it does use sentence and text, right, to generate the text without bugs.

[03:03 - 03:12] Another perhaps slightly obvious thing is to avoid generic language. So what do I say when a generic language?

[03:13 - 03:22] Things that don't use the active voice, like instead of saying things like the project was delivered, right? That's past, right?

[03:23 - 03:30] You wanna do something active, like, I deliver it to the project. Right, you're relating with your subject, but you and your verb.

[03:31 - 03:50] So I verb, or sometimes you can abstract that the pronoun and say delivered or use the infinitive when you're talking about responsibilities for all, delivering XYZ when you're speaking more generically about the function of the role. But generally avoid active voice, use strong verbs.

[03:51 - 03:57] Weak verbs are things like worked, cooperated. If I see those words, that doesn't tell me anything.

[03:58 - 04:04] Okay, you could have been bringing coffee to the team, that would have been worked, right? That would have been cooperated.

[04:05 - 04:25] I'm talking things like delivered orchestrated, built as a good one, automated, stuff like this, right, optimized, these are your strong verbs. To look at an example, a weak example, I worked with teams in the AI space that worked on SEO integrations.

[04:26 - 04:31] That's quite weak, right? We have worked twice, which really highlights how weak of a verb it is.

[04:32 - 04:38] But this really doesn't tell me anything, right? It doesn't tell me about the copies you use, it doesn't tell me about the impact you had.

[04:39 - 04:43] And I am going back to what I said about being quantifiable. Let's use a better example.

[04:44 - 05:05] I developed a strong verb, product strategy, and ramp a plan for two offshore development teams building AI analytics solutions for SEO layer, right? The same, a lot of the same keywords that are being used are you have AI, but we're using AI analytics instead of space, that's more specific.

[05:06 - 05:13] We're talking about SEO, that was sort of core technology here. But we're also giving some quantifiable stuff, right?

[05:14 - 05:22] We're saying the project was completed three weeks ahead of schedule. Maybe unrealistic, I haven't worked on many projects that are delivered ahead of schedule.

[05:23 - 05:34] And the ones that have don't always tend to be great, but this is just to give you an idea, right? Like you have a really clear time and scope here, and you have things that are measurable, right?

[05:35 - 05:40] We are even saying that two offshore development teams, right? We're embedding a lot of information there, right?

[05:41 - 05:52] You're saying that your scope was over more than one team, you're implying some kind of stakeholder management, some kind of collaborative work. And the fact that they're offshore also gives a lot of context, right?

[05:53 - 06:01] It shows that you can work at this kind of distance. And you're achieving all of this by using very specific language that have a lot of information embedded into it.

[06:02 - 06:10] Next, format. And again, this is very much in with the fundamentals, right?

[06:11 - 06:29] When we talked about consistency, clear headings, simple formatting, distinct boundary between the types of information, lots maybe more important for human readers than for LLMs. But nevertheless, it's good to lay out your CV in such a way that it can be easily understood and easily summarized.

[06:30 - 06:39] You should use clear titles, right? So skills is probably better than developer skills overview, which is better than my awesome skills.

[06:40 - 06:51] I've never seen someone go as bad as my awesome skills, but it wouldn't surprise. But yeah, being clear, being somewhat predictable, at least in your layout is useful.

[06:52 - 07:01] I also recommend a target-length for a CV of something like 700 works, 700, 800. Generally, you're looking at a bit two A4 pages.

[07:02 - 07:05] And that's why you need to be so specific with you. You lose use of language, right?

[07:06 - 07:18] Every word counts, every word should be demonstrating some kind of skills, some kind of value, some kind of demonstration of why you are a good hire for a role. A brief note on document types, since it starts to format.

[07:19 - 07:28] There's a lot of different document types out there. It's very attempt to use something like Latex, but can make these obstacles, the gorgeous documents.

[07:29 - 07:34] But generally, use a word doc, just use a word doc. That's no need to reinvent the wheel.

[07:35 - 07:43] Word docs are going to integrate into the ATS systems that companies have. But you want something that's widely interoperable.

[07:44 - 07:49] It might be going from one software layer to another. So you want something that's going to stay consistent.

[07:50 - 08:09] PDFs are probably okay. But if you have a complex back end, there is a good chance that sometimes some kind of screening layer, some kind of extraction layer will fail and you might end up with some noise going as an input into one of the screening tools.

[08:10 - 08:15] Some hiring sites will actually prefer PDFs. And in those cases, do use PDFs, right?

[08:16 - 08:25] But generally speaking, you should use a word document, just on the off chance. I also want to talk a little bit about cold outreach.

[08:26 - 08:32] So what is cold outreach? Cold outreach is reaching out directly to recruiters.

[08:33 - 08:43] Now, this can be a really useful thing to do, especially if you're particularly well-suited to a row. But approach this with a lot of caution, right?

[08:44 - 08:55] It's really not always appropriate to reach out to recruiter, especially if you're applying for quite a lot of jobs, especially if you haven't super tailored your resume. And again, this is really a contextual thing, right?

[08:56 - 09:13] Those recruiter give a signal that they're open to being reached out to, right? I see some jobs on LinkedIn or on Indeed or whatever where you can apply for the jobs to Indeed, but there's also a note that says, to apply for this job, send an email to Yara Yara, right?

[09:14 - 09:25] In those cases, they yes, absolutely do cold outreach. Honestly, if it says to reach out to the recruiter, it's not even really cold art outreach at that stage, warm outreach.

[09:26 - 09:44] But when you are approaching a job like that, to put more effort into it, right? A cover letter mentioned, especially things, you know, that might not be evident from your CV that make you ideal for the role, maybe brag about a project that's in the same space, stuff like, obviously, be polite, be respectful, be friendly.

[09:45 - 10:00] And don't do this for roles that you don't take your particularly well-suited for if you are casting a white net. Okay, so I'm coming back to tailoring your resume because we did talk about this before, but it bears repeating.

[10:01 - 10:18] So the level of tailoring is going to depend on the kind of role you want, right? So if you're applying for a lot of jobs, tailoring your resume to each and every single one, that's got a high opportunity for us, right?

[10:19 - 10:23] That's a lot of work. So you want to be smart about this.

[10:24 - 10:35] Obviously for roles that you think you have better chances in or that particularly want, it's worth it to put in the extra effort. It's also worth noting that a lot of people aren't just looking for one kind of role.

[10:36 - 10:58] In my case, I've worked as a QA engineer, I've worked as a technical project manager, I've worked as a technical writer, there's other roles that I'd maybe be interested in but might not have as much experience for, but I obviously can't use the same template for those roles. If you're applying for different job palm leaves, different levels of similarity, whatever, it's good to have different templates for each of these.

[10:59 - 11:17] So you can quickly do a little bit of tailoring, a little bit of fuddling with keywords or whatever, for the jobs you apply without spending a huge amount of effort. Again, you can't always include every, it's information, especially if you've had a very career.

[11:18 - 11:29] But I also kind of want to talk about the strategies for what level of tailoring you should give your resume. But I'll come back to that after this, the tools and platform section.

[11:30 - 11:45] So we talked earlier about how some ATS systems will scroll professional profiles. This is stuff like LinkedIn Indeed Upwork, if you do freelancing, your GitHub repos.

[11:46 - 11:55] So it's quite important to keep these up to date. We did see that like 40 something percent of AI screening tools will search these, right?

[11:56 - 12:03] So definitely good to keep these up to date. Automated resume checkers might be able to follow links that you have in your resume.

[12:04 - 12:11] I usually have a link to my LinkedIn profile embedded in my resume somewhere. After some more code heavy role, I might also include a link to my GitHub portfolio.

[12:12 - 12:19] Stuff like this will be checked by automated resume checkers. On minimum, they might check if it's valid.

[12:20 - 12:23] So don't include broken links. Make sure your links work.

[12:24 - 12:28] But yeah, it's worth to keep all of these things up to date.