Episode 40

Selvaraaju Murugesan on The Future of AI in Hiring: Efficiency, Clarity, and Innovation

TIM: So thank you so much for joining us on the objective hiring podcast.

SALVARAAJU: Yeah, thanks, Tim, for having me.

TIM: It's a pleasure to have you here, and I would love to start with a theme that's probably on everyone's minds and lips: AI, in particular. I'd like to go into AI and hiring and just ask you, have you had a chance to dabble with any tools in that AI hiring process? Any kind of steps of the process that you've tried to use a bit of AI to improve?

SALVARAAJU: Look, the AI comes into play when we have a lot of applications at scale. Basically, if you have one or a few applicants applying for a role, then it's very easy for a human to actually go through the CV and then assess and then filter out, but as the volume of applications scales, in our case it actually happened at one instance when you are hiring for an intern role. more than a few hundred applications coming in, so it was very hard for our HR process to actually filter it through. Then we started looking at tools like GoHire and Fresh Teams; those are the ones that we use in the organization. We also look at Zoho Recruit, and they're really nifty tools in terms of how we can actually filter out the applications. So we pretty much use some of these tools in the pre-screening process in terms of filtering out the applicants before it comes for a human evaluation, and it's been good. I In terms of how much it can actually filter through, you put in pretty much all your prescreening criteria in terms of what you want to look for and how you want to help the AI to actually pick the right CVs, and then based on your criteria so it's very important You set the criteria first, and then the most important thing is, you know, To know about all this AI hiring tool is that you have to be very explicit and clear in terms of what you want as a selection criteria. If you're a little bit ambiguous, then obviously the AI will go haywire, and then it'll probably pick a lot of candidates or probably a few of them. So it's not useful at all, and then it also actually helps to improve some of our processes as well in terms of how we prescreen people in the old business process, but with the new AI tool in place, it actually completely changed how we do things here.

TIM: How did it change how you do things? I'm interested.

SALVARAAJU: So in terms of the efficiency, if you look at a normal HR going through your CVs of a lot of applicants, obviously at least they have to spend quality time; in our case, if you're hiring for a data role, especially for a data analyst or a data scientist role, there's a lot of information they have to go through in their CV. For example, what kind of tools they use, what kind of projects they've used, and, more importantly, what kind of outcomes they actually produce for a company—those are very key things for us in HR to actually go into a CV and then get their information out. So at scale, it's extremely hard to spend a quality amount of time if you spend five minutes on one applicant, then obviously if you have more than 500 or 1000 applicants, it just goes. It's not an efficient process to run through, especially in HR, and especially the business is demanding to hire us a data scientist in a few weeks time, and they have to go. Get ready for a project to go live in a few months time. I don't think it doesn't actually fit into the organizational agility of things, and it's brilliant that this AI tool, especially, is helping in HR processes.

TIM: Yeah, I think these tools could really help us rethink our expectations over how we hire people because I feel like the way traditional hiring has been done has been pretty slow. The fact that two weeks would be considered a fast amount of time to hire someone, I feel like we could certainly chop away at that and the amount of busy work in hiring as well, like reading CVs, manually scheduling interviews, and sharing feedback from an interviewer to a candidate. There's so much crap basically in there that surely if we could automate away some of this stuff, we'd be able to spend more quality time with candidates.

SALVARAAJU: Absolutely, Tim. The more important thing is if you work in a SaaS company, we're in a software company where business agility is the key to winning the market to get the business advantage, so most of the founders and the CEO are very keen on making sure that we hire the right candidate in a short span of time. So that's mostly the messaging they give to the HR team. You had to hire the best candidate in a short span of time, and there's no compromise to this, okay? And that's why this You have to completely go and reengineer your hiring process to help your organization be a little bit more effective in terms of hiring at the same time. Look, the moment you hire the talent, what do you want to do? You want to mobilize the talent; you want to put them in projects or work on a product feature or things like that so that they can actually get going and then build that ship, the feature, and then this is this value delivered. So that's the complete kind of cycle they're looking at, and in that role, definitely the AI tools help.

TIM: Yeah, you make an interesting point there, and I feel like it's something we gloss over. If you're so in the weeds of hiring, sometimes you forget to realize, Hang on, I'm hiring someone to do something to add business value. I'm not hiring someone just for hiring's sake, and so you get hung up in the actual day-to-day of doing the hiring steps without realizing that, like, I need them to deliver value, and they can't deliver value until they're in the seat and the right person

SALVARAAJU: Absolutely. Look, there are many scenarios in the company. For example, some of the stakeholders in the company, they'll have a budget allocated, and they don't spend it. They don't get it in the next run. Obviously, they will hire just for the hiring's sake, just to have one member, somebody in the team, do some mundane task, but they're not actually adding value. and that's not how you should hire. Basically, if there is a strong need, and most importantly, they should deliver value, that's why we hire talent for apps.

TIM: Startups or scale-ups normally have those kinds of pointless roles. Maybe that's a slightly more corporate thing where there's that inherent inefficiency of a massive organization that sometimes creeps in but probably not in a 10-person startup

SALVARAAJU: Oh, absolutely. The startups are actually the gold standard for this hiring. The way they hire, they invent a lot of unconventional ways of hiring as well, so I think this space is definitely different, interesting, and close to my heart because for some of the roles we don't advertise, but we go and then look for people in conferences, meetups, and some other ways where we actually hire talent through that. Not through the regular, yeah, put the CV, then this HR process and stuff like that. There's a lot of startups that actually go and innovate in that HR hiring space as well.

TIM: You mentioned something in passing that I thought was really helpful for people to understand, and so you said you have to be really explicit and direct when you're setting up the criteria for the AI to follow, and I thought, Yeah, that's true; that rule also applies even if you're doing it with humans, and I feel like part of the reason why traditional hiring has failed So often, maybe we haven't thought of exactly what we want from the get-go, and we're making it up as we go along. So the fact that you're almost forced to give really good instructions immediately is almost creating a better process as a result as well.

SALVARAAJU: Oh yeah, absolutely, absolutely. Look, it also brings clarity in terms of hiring; more importantly, if you define the job, especially the role and, most importantly, the responsibility of what the person would do, it actually aligns with—look, it's a two-way transparency. It also helps the employer in terms of setting up the expectation of what the applicant would do. and also from the applicant side, they are very clear: Hey, is this matching my skill set? Can I deliver? Have I done this? Have I delivered this role and response in my previous employment? Is it aligning with my skill set? If that's the case, then they can actually apply; otherwise, don't bother because it's not just humans or AI in the loop. During the prescreening, you got to be very clear in terms of what you can actually deliver, and that's very important. In fact, I would actually go and say that having an AI in the whole process brings a lot of clarity to the employer in terms of setting up the clear expectations. In terms of roles and responsibilities, and more importantly, in terms of what the day-to-day activities look like, they have to explicitly say that and what kind of business impact they are actually looking for in the applicants, and if it's there, then it's very easy for a human or even an AI to go through thousands of applications at scale with a few seconds, and then they actually pre-screen the applicants and then give it for human evaluation in the next step. yeah

TIM: Yeah, and it's interesting now to think about like old world versus new world. Some companies, if they were really strong at hiring, they would think very carefully about these things ahead of time and think about all the kinds of questions that a candidate might ask in that first interview and almost bring that to the front of the process, but the challenge was always that it was unscalable; you can't share all that information easily with candidates, but now with an AI tool, maybe sitting at the top of the funnel, it could answer a lot of those questions for candidates, maybe to cover off a lot of those things they're interested in finding out.

SALVARAAJU: Yeah, absolutely, Tim, absolutely. Look, it'll be very interesting. I'm pretty sure in the new world there'll be like an HR chatbot where the candidates will actually have a conversation, and then it will prescreen based on that conversation, and there are tools we actually tried where even for the interview process you just ask a lot of questions and you have your criteria, and then actually it writes the feedback for you. So there's no human feedback, and then you, the employer, we have to do is okay, read through those notes. Yep, it aligns with what I want to say as feedback, and then within a few seconds, once the interview is over, you can actually give feedback to the candidates, which is helpful, and if you do not want the candidate to go to the next round, it's also reducing the effort required from the employer side as well. So it's a win for both the parties involved, yeah.

TIM: Yeah, and that's a great point there around feedback, and that would probably be a candidate's single biggest complaint: getting no feedback or getting crap feedback, so the fact that these tools are solving that problem is amazing, and I think it's also like I, over the past few years, felt a lot of almost hesitation from HR and talent teams around AI and the perception that it's almost dehumanizing or somehow going to have a worse candidate experience. This is a great example where the opportunity is to have a much better candidate experience than doing everything manually

SALVARAAJU: Yeah, absolutely. The most important thing is, and there's a human in the loop, obviously we cannot rely on the AI to write all the feedback, especially we, the As, because the human is actually giving feedback to the AI, so the humans are emotional beings. So whenever you see those very hard comments and stuff at the As, you have to have a human in the loop, so once the AI writes the feedback, then the HR or the employer actually reviews that and then sends it and then probably rewrites a little bit, but it actually saves time definitely at scale when you do it. It's much more efficient, especially for the larger organization hiring for a role where there's a lot of demand. For example, when you hire for a data scientist, an ML engineer, or an ML Ops person, there's always a demand, so you're looking at least thousands and thousands of applicants pouring in their CVs at you, and you definitely need to have a tool in place to prescreen. At the same time, you also give valuable feedback because that's how your brand gets recognized, correct? What if you're a startup and then you're not giving any kind of feedback, and then the word gets out to people? You know, you go to all these forums, and they talk about your company, how the hiring process is broken and stuff like that. Which actually creates a lot of brand damage. So to be sure that your organization has a good brand reputation, you must provide feedback to the candidates after the prescreening process. and that's a human way of doing things as well. There's a humility in the employer as well. Look, you applied for a role; we value your time; we want to give you some feedback so that we actually improvise, and then I get a job, if not in this company, probably somewhere else. So that's a good intention of the employer side as well.

TIM: Yeah, and for candidates who've been used to getting no feedback consistently, now some companies are going to start providing it, and then that will probably force other companies to lift their game. There's going to be this competitive element to it because, as you say, otherwise the candidates are going to be irate and start burning them on the online socials pretty quickly.

SALVARAAJU: Yeah, look, these days once a candidate comes out of an interview, they immediately go to Twitter and then tweet about it. It's that crazy, and then a tweet can go viral in no time.

TIM: What about then? So at that application stage, let's say the AI is now getting involved to do that more objective screening; that's obviously going to be a lot more efficient and faster. One bit of feedback we've heard quite a lot in the last few weeks is that the CVs themselves tend to be getting more and more similar looking to each other, and they seem to be getting better in the sense that they better match the job description. Presumably because candidates are using a large language model to optimize their CVs, what some people are saying is, Yeah, I've got 500 CVs, but 400 of them look good. So even if I could automate the screening, which one am I going to choose because they are all amazing looking? Where do you think this is going to go then? like what's going to be the next step if all The CVs are amazing.

SALVARAAJU: Yeah, look, it's very interesting. The first thing we actually do before we put the CV is whether the CV is generated by a chat GPT or LLM model. If it is, then we don't anchor it unless there is, so we have to look for authenticity. So even a lot of these emails and stuff like that, we build a lot of plugins where we actually check whether the current is actually generated by a chat GPT-like interface using large language models. and I've seen in many applicants where they put in a lot of these keywords because before this AI, this large language model, there was a technology called natural language processing, so it actually screens based on keywords, so I've seen people using certain keywords from the roles and responsibilities and the skill set in the CV so that there's an exact match. Then the prescreening happens automatically for them, but these days the LLMs are different, so, as I said, when we look for any kind of a job applicant, we actually set out. They have to write, in addition to the CV, what kind of projects they worked on and what kind of outcomes they actually produced. and they have to narrate a little bit of day-to-day activities, which they cannot cheat; of course, they can actually use an LLM to write that even though they haven't actually executed that role, so we will find out because the content is actually screened through a process where we actually check whether the current is actually generated by chatGPT. If it is, then they're not the right fit because they're cheating from the first place, which means they are not apt for the right company.

TIM: And so that's a philosophical question, so it sounds like you would consider using ChatGPT in the hiring process as cheating.

SALVARAAJU: Not as cheating but as a way to when it's cheating if they haven't actually done that role, how could you ask a chatgpt if they haven't actually delivered the project day-to-day activities and stuff like that? You can't ask chatgpt to make up things that you think that you've actually done, okay? That's wrong. that's morally wrong If you have actually delivered that role, then you can actually articulate the day-to-day activities, and for day-to-day activities, it will be different for all the candidates; it will be different for different candidates because the nature of the work they do and the toolset they use are all very different. So then we'll find out we also started doing analytics as well, but our HR also does a lot of analytics on those prescreenings, how much our chat GPT generates, but we find that a lot of people are genuinely writing CVs again, the CV hiring from the CV. I don't believe in hiring from CV again. Tim Okay, I look at it cool; I can look at your LinkedIn, and that's as good as your CV. Okay, show me the projects or show me the work or any kind of projects or features that are actually shipped that actually produce impact for the organization, then if you can actually articulate that, that's fantastic. Then you know how much you've actually worked and how much it has been put into use by the company, and there's a lot of value delivery happening if you're doing any fundamental research. We also want to look at what kind of research paper you actually produced and how much of the new knowledge was actually created. And that's very important, okay, to showcase that kind of a talent or skill.

TIM: I wonder then if where we will go is if there's a system to because I feel like part of the problem at the moment is any data that a candidate could put onto a CV isn't that valuable because you can say anything. It's not really validated at the point at which you've written the CV. Like nobody can cross-check that; you have to take the candidate's word for it. They have to take them at face value, but I wonder if there's a way that we could start to unlock new data sets, for example, what the candidate has actually done in their jobs day to day. I don't know performance review data sets or task management data sets, or, as you said, maybe publications or GitHub repos, and because all of this stuff is unstructured, maybe that would be a good use case for a large language model to aggregate all of this stuff. what

SALVARAAJU: Yeah, yeah, yeah, absolutely, team. Look, you're spot on because, in fact, we actually hire based on their GitHub profile. No need to send this CV. In our organization, we hire only based on GitHub. Okay, the reason is you can't lie in the public forum. Okay, if you forked or cloned a project, if you copied somebody's works, we will know. How much code contribution have you given to a public project? How many new projects have you worked on? How many collaborators do you have? How many code comments have you done? How many ground-up innovations have actually been done? And most importantly, we can look at your code; we can look at how much coding stands have actually been used. Have you used a variable named ABC? For example, okay, we shouldn't use that, but we look at whether they actually had code comments, whether they have a good Git commit process, and stuff like that. The more you dive into this, okay, then you clearly understand whether they are the right fit for your organization. In fact, we have a proprietary kind of code where once you give a GitHub profile, it actually goes and analyzes that GitHub profile and gives us a report. So I get a report when it actually analyzes it. It does it in 5 to 10 minutes. It gives me a report on how many collaborators and how many projects. How many codes have been plagiarized? How many are new novel codes that they actually added? What's the frequency of the code comment? How active are they? So it's very important for us to actually know that, and we also look at how many lines of code they actually produce in a year. Okay, basically we have a metric that each programmer or software or data scientist writes about 10,000 lines of code every six months, so we look at how much code they have actually written. Are they building any applications and stuff like that? Yeah, so definitely there is a You can't use CV in the AI process; definitely, you have to have a more unstructured way, as you said, Tim, in terms of Hey, tell me in two pages about a project that you worked on, okay? And then you actually put a lot of criteria. Let the candidate write it. Okay, then you put it through the HR tool, and then you can actually evaluate it better and prescreen it better. I think the CVs are very old paradigm. I fundamentally believe that the CVs will no longer be applicable in the next six years or so. Okay, once the GNI becomes a lot more dominant in terms of adoption, then definitely CVs will be outdated.

TIM: I'm interested in this GitHub analysis you've done and what the origin of that was like. When did you first come up with that? When did you decide to scale it into an actual script that would calculate the metrics automatically? I'd love to hear more about that.

SALVARAAJU: It started four years ago. From our organization, we started receiving a lot of applicants for a data scientist role. There's a plethora of them, and it becomes very hard for our HR to actually screen them, so then we looked at what could be the best criteria to evaluate because the candidates can lie about whether they have a skill or not. In fact, some of the prescreening things we actually found out were that somebody else, well, there's a proxy for the candidate, and they will take the prescreening questions, and it's all been happening, and then we've been gathering data, and then the candidates are using any effective methods to cheat the system, cheat the process. but with the GitHub profile, they cannot cheat; it's a public profile, and what they got is what they have, okay? And then we realized that, okay, even though people started putting GitHub profiles in their CVs and then also in their LinkedIn, then we started looking at those GitHub profiles, and what we found was what the candidate had in the CV. It's not reflected in the GitHub if the candidate says they have done something like Node.js or Python or stuff like that, but their GitHub profile says they've done a lot of C++ projects or Java projects. Definitely, they're not; they're lying. Probably they're learning, but definitely the GitHub profile is not reflecting what they're actually saying in their resume. So then we started, okay, we look, we need to automate this because we have a small HR team inside our organization, so we want to scale the process. Then we started looking at how, from the data science team, that's our own ground of innovation, how do we go and assess all this GitHub profile? So we started writing a lot of scripts using Python, and then when you give a small get a profile URL, it actually goes and pretty much scans their code and looks at all the repositories and gathers a lot of information about description code, and then we run our own criteria on the top of it, so we get the analytics basically. and then those analytics are the ones that get to the HR hands, and then they pre-screen based on that, okay, and then that's how we actually go from thousands of applications to a few hundred applicants where we can actually go for a human evaluation.

TIM: What an interesting approach! I wonder if you ran into a couple of issues with this one. It might be candidates that don't have a GitHub profile; maybe they use another repository, Bitbucket or something, and if so, how did you

SALVARAAJU: No, yeah, it's not just GitHub; any Git repo for that matter, Tim. Yeah, it could be Bitbucket, or it could be something else in the open-source world, but Git is GitHub is predominantly used by most of the developers around the world. Look, if they do not have any kind of Git account, for example, they do not have any kind of public project company of any projects in the public domain. Look, they are not a great fit for our organization. Yeah, it's very simple. You are hired as a data scientist and analyst. You need to know how to program and then write software basically, so if you do not have any Git profile, yeah, any kind of Git profile is probably not the right file.

TIM: So it's almost like an implicit test in itself.

SALVARAAJU: Yeah, absolutely, Tim. Look, the other important thing, Tim, is a lot of candidates we've actually seen is they're using web upload, like they go to Git and then they upload a zip file or something, and then we do not want them. Okay, we want people to actually use Git commands because, come on, if you work in software, nobody's storing a file. They have to do version control, so unless they know Git commands, they're not a right fit. When they use web upload, that's the first red flag we get from that tool. Yep, then the HR will say, Look, yeah, they have a GitHub profile, but they're not the right fit because they are not well versed with the GitHub version control.

TIM: the kind of technical versus soft skill trade-off I wonder if you had any thoughts there, especially with where you might see large language models going. For example, maybe now compared to three years ago, the value you might place on writing SQL or Python from scratch is different because now a large language model can do some of the basic work for you, so I'm wondering if you've had a thought about that, yeah, kind of the soft skills/technical skills trade-off.

SALVARAAJU: Yeah, definitely the technical skills are very important. Unless you know how to use the programming language to solve a business problem, you're not going to be better at doing your job in terms of soft skills. It's very essential. For example, some of the candidates we interviewed, they are technically skilled, but they have communication problems. For example, they're not good at writing, and yeah, in that way they can put a prompt in ChatGPT and then make a few hints and actually write something very formal and very professional using the right words and grammatically correct sentences, and that's good, okay, but that's great. Writing emails is one part of the job, but more importantly, as a data scientist, you have to go and explain your algorithm or present some data insights to the C-level executives. and you can't use an LLM for that. Okay, the most important thing is the communication is extremely important, so we look for more in terms of storytelling skills, whether they're able to engage the audience, especially the CLL executives, using very simple terms to explain complex data science concepts or the algorithm actually built and how it can be put into production. So we value a lot more in terms of storytelling skills, okay, and then that culminates everything from visualization using the right kind of vocabulary and using the right words to actually express things and stuff like that, and then the other soft skill apart from storytelling is collaboration. As a data scientist, you have to collaborate with data analysts, data engineers, business stakeholders, your software engineers, your testing teams, and the documentation team, and it goes on. Are they the team player? Okay, it's extremely important we test those kinds of soft skills in terms of how collaborative they are. How much persuasive they get when they have to push through their agenda or the things into certain things into certain places, we definitely know there is a trade-off between technical and soft skills; definitely you cannot have a perfect candidate where you have their perfect They have exactly technical skills and good soft skills. It's extremely rare, but that's when, in the organization, we have a lot of training programs. If they're technically sound but they have a lack of soft skills, we provide ongoing training to improvise so that they come to the organizational standard in terms of delivering excellence at work.

TIM: Do you have any views on whether or not it's easier to upskill someone in the soft skills or the technical skills? Is one easier to teach or one easier to learn, maybe?

SALVARAAJU: Yeah, look, if it's a binary question, I would definitely go for choosing them to upskill in technical. The reason is that technically, it is very easy when you have a programming language or a new kind of paradigm coming in, like a large language model. It's easy to teach those technical skills, but soft skills take longer, and the reason why it takes longer is you have to practice. A lot And if it's natural in the candidate, then it's fine, but if it's not, then it definitely takes longer time, so I would prefer a more soft skills person than a technical person in my perspective because we can definitely have a training program to upskill them in technical skills, but soft skills definitely take longer time. And if you look at the returns for the investment, if you put it on the technical side, the returns are very quick, so if you teach them Python or Node.js or Java and anything like that, within a few months they are good at it, and they can be productive, but if you provide soft skills training, it takes a longer time for the organization to get the ROI back. and also for the candidate to practice it actually put into their own behavioral change and then become part of every day-to-day activity, it just takes longer time to settle in, so definitely my take is technical skills; yeah, we can actually upskill you, but soft skills Yeah, I would prefer to have a more soft skill person than a technical one.

TIM: Yes, and I feel like, from my end, an important caveat would be that for the technical upskilling, I'm assuming this is on the basis of an already strong foundation. So, like teaching someone who already is a programmer to learn another language is easy, teaching someone who has no concept of anything analytical or programming at all to teach them Python would be very hard. I would have thought probably as hard as teaching someone with weak written communication skills to be a good writer.

SALVARAAJU: Yeah, yeah, absolutely, absolutely. Yeah, I think the world is looking at, given the large language model, is anybody even giving a small prompt? Okay, yeah, yeah, yeah, yeah, very subtle, and then ChatGPT can write you a marvelous piece of content for you, okay, very professionally written, so I think helping technology is helping people to write better. But again, when you are in a board-level meeting where you have to communicate that Child GPS can't help you, you have to be very authentic and very articulate in terms of explaining concepts, and then more importantly, how do you win the argument, or how do you make sure that you push through your insight? so the mistress so that the stakeholders can actually make effective decisions, so those are very crucial in terms of when you hire for a data analyst or data scientist role.

TIM: I wonder if even that will change in the next few years. Some of these kinds of Google Glass technology and neural links must be within 10 years, maybe, of having a direct connection to a large language.

SALVARAAJU: Yeah, look, to be honest, when you took a look, I went into a lot of this data science conference and stuff like that, and the moment I came out of the DA conference, I actually felt what it's like to be a human, okay? The real fact is the A is taking over a lot of things as a human we value a lot. okay, for example AI came for labor, okay? For example, running trains, you have autonomous vehicles and things like that. It came for most of the mundane tasks we do, but most of the people are valued because they can think; they can reason at the moment. The technology, especially the AGI, the Advanced General Intelligence, is actually testing what it's like to be a human. What value add am I giving to the world for being in this world? Okay, it's a very philosophical question, and it's very hard to imagine that world, okay? And if the technology is going to be there in a few years time, yeah, it's going to break a fabric, yeah, of our society. and we just do not know how it's going to be. There's a lot of doomsday scenarios and stuff like that, and then that's what I don't need to do. I don't become anxious thinking about it, and the more important thing is even if there's an AI that can think like me or think even better than me, I have to differentiate. Okay, that's more important, so I focus on that deep-level thinking rather than competing with some kind of intelligence that's artificially created.

TIM: Yeah, and I guess in the meantime we can leverage these tools to our massive advantage, especially given we work in data, and we're at least maybe not quite the bleeding edge, but it's somewhere near the forefront of understanding how these tools work, so we've got to be at an advantage.

SALVARAAJU: Yeah, absolutely.

TIM: right now

SALVARAAJU: Yeah, absolutely. Yeah, yeah, I use a lot of AI tools in my day-to-day activity, so that actually helps me to be a lot more productive and absolutely

TIM: What about thinking now around sourcing candidates, like you mentioned earlier in the call, about sometimes going to meetups and having slightly unconventional ways, especially as a startup, to try to get candidates as opposed to the usual job ad on Seek or LinkedIn? read CVs, so can you expand a little bit on some of those other techniques you're talking about?

SALVARAAJU: Yeah, I do a lot of—I attend meetups, and I also press in a lot of meetups, so when we present on a topic and then we look for a lot of people, and then we really want to understand what's the motivation, okay? Are they looking for some kind of explanation or queries on your presentation, or are they curious in terms of learning new things? And that's one important thing, and then once we know a lot of people, familiar faces, where they are actually working and whether they're actually looking for opportunities and stuff. and then we also see a lot of new people coming into the meetups, new faces who are also attending the meetup just to get a job, just to land on the job, and we started doing meetups and then going to conferences and presenting. We look for people who are naturally curious and, more importantly, they are skilled enough. We also look at whether they have the intent to actually attend that meetup; it is to get a job. They could be attending that meetup for the first time, and they see me presenting, and then they'll say, Hey Selvar, do you have an internship role, or are you hiring for a data scientist role? If that's the thing, we do not want them. Okay, we just avoid them because they are very outrageous in terms of asking that open-ended question because you haven't actually proved anything to me. Okay, so the way we hire through meetups and conferences and other kinds of events is we look for a lot of familiar faces that are coming often. and even if they don't have a job, that's fine, but they have to come regularly because that means they have an inherent characteristic of upskilling them, keeping up pace with the technology; they want to know what's actually happening in the domain, so that actually shows there's a curiosity, there's an initiative they're actually taking in There's a self-improvement process they have in place. Good ethics to have, and then we'll have a conversation with them in three to four meetups when they see them. Hey, are you? What are you doing? Show me your GitHub profile. What are the new projects you're working on? And then we make some kind of an assessment, and then we go through, and then, hey, look, there's a role that's not advertised, but we are actually hiring for the troll. Would he be kind enough to be considered for the troll, and then that's how we approach those kinds of meetups and events? And for candidates who are actually looking to be hired, I think it's the best way to go to conferences and meet a lot of people and then network with a lot of great minds in the space, especially in the data space and AI space. I get to know them better. but you don't have to be a little more ambitious in asking them. Hey, can I get a job in the first meet rather than take your time? You have to settle things down a little bit and then show curiosity about projects and then let them showcase and let the project speak for itself. Then let the employer or the person who wants to hire you be interested in you, and then you can actually make a move. I think it's a long-term strategy, and that will work out if you have only targeted looks. I want to work for Atlassian or Commonwealth Bank or any kind of bank of that sort. If you're very focused, you're going to go and do these kinds of things for the long run, and then that actually sets you in a stage where you can get hired, and then the chance of getting hired is more

TIM: Yeah, it's a great distinction that you've made there because going to a conference or a meetup Hey Selva, can I have a job? is the sort of real-world equivalent of sending someone a LinkedIn message. Here's my CV. got a job like It comes across as very aggressive, in-your-face, and very transactional, and I think it's going to put everyone off who experiences that, but then you've just outlined this much better way to think about it where it is a long-run investment; you're building a relationship; you're showing your value; you're showing your worth, and you've almost flipped the equation on its head where now the employers are looking at you as opposed to you begging the employer for a job.

SALVARAAJU: Yeah, absolutely, Tim. Look, that's the other unconventional way to get hired. To be honest, I've seen people sending me LinkedIn messages: Hey, here is my CV; get me a job, or Is there an intern role? Yeah, normally it's very tricky. Okay, we get thousands; we are flooded with thousands of those messages every day. The most important thing is, okay, so if you follow an employer, you can have a lot of open communication; for example, they will have the corporate plan or corporate strategy out there, and then if you follow certain people who are in your space, for example, data science or AI, they also do a lot of social media. They present on some of the topics they have done their projects and stuff like that, and then what the candidates need to do is pick up a problem that they know and then try to solve the problem using a GitHub profile and then email that person. By the way, Selva, here is something I found you presented on something. Here is an improved thing, and then when I get an email, I'm intrigued; I'm already hooked. Okay, somebody is out there. I haven't actually paid them, but they actually have a vested interest. They build a GitHub profile, they build a project, they have a report, and they walk me through how we can actually solve the problem. They're actually adding value to me, to my job, and then obviously I'm naturally intrigued. So obviously I want to have a conversation with the candidate doing an outbound in terms of hiring. It is very effective, and if you do it very well, you can land on a job in no time. A lot of this is extremely against the normal hiring process. Emailing someone, say, an employer, with a project that you worked on for a few months and then with the report takes time; it takes guts, but it will definitely land you on a job. It's very rare to see those kinds of things, but if you execute it well from the candidate applicant perspective, that shows that you have a vested interest in the company, in the person in the job, and you're actually showing a lot of these cultural traits: you're initiative, you're self-driven, you're motivated. Yep, great, and I definitely want to have a conversation with you.

TIM: yeah I mentioned part of the issue for candidates is often that they'll be applying to various jobs through jobs boards they'll then know and realise the market conditions are against them in the sense that there are so many other applicants they start to panic especially if they're out of work or they're a recent grad or something and that they just need a job so they go okay I should apply it to more jobs so use this kind of spray and pray method which then dilutes their effort and focus and then they apply almost the same strategy but to this networking approach and it's never going to work because of what you just outlined because it's going to be too generic too in your face too transactional whereas if they maybe had this targeted approach where it's here's the three companies I want to work for how can I get a coffee with three people from these companies that is quite possible but you have to do it in a nice way

SALVARAAJU: Yeah, spot on, spot on, team, spot on. Yeah, yeah, I really like that your phrase, How do I get a coffee with this person? That's a very good conversation starter. Okay, if you can, that could be the good outcome for that outbound activity if they're able to send some of the things, and if the employer responds or the person responds back saying that Let's catch up for a coffee. I think that sets the stage for you to land a job, but again, you can't be aggressive; you're going to take the soft approach, take one step, and then build the trust with this person and with the employer, and then in the long run you will end up there definitely as one of the candidates or one of the people who pretty much land a job there.

TIM: Yes, and again I feel like the nuance here is very valuable for candidates to understand, so it's not Oh, a candidate knows they just want a job from you so far, so they come up with some bullshit reason to get a coffee with you, and then as soon as they see you, Oh, by the way, here's my CV. Can I have a job? No, no bait and switch it has to be like an exchange of value, it has to be real, okay? There's no lying involved; it's just going into it without any expectations. You're going to build your network. Why wouldn't I want to have a coffee with Silva? Anyway, it's a good person to have in my network. Maybe a year later, maybe 10 years later, something might come up, but viewing it transactionally is going to be a straight ride to disappointment, I think.

SALVARAAJU: Yeah, that's right. Look, if anybody's listening to this podcast, I think that's a great message for them, given this new Gen Z/Gen Alpha generation. They're very transactional in nature. I think for them it's a very old way of doing things, but it just works because that's how the human sees it. And most of the employers see it, and they definitely make sure that the candidates have good ethos, and then once you hire the candidate, they're in; they should be with the employee for the long run because the employee is also investing in the applicants or the employee. They definitely want to hire the best that suits your organizational culture, and at the same time, they're actually adding value as part of their role.

TIM: One of the things I thought of just then for candidates is if you're going to use this kind of outbound approach to target particular people that you'd maybe like to get a coffee with or network with. You know the process that any sales organization would go through is identical; like, it's a funnel of people. You identify a set of accounts or companies you think, Who within this company would I like to speak to? Why would they speak to me? What's my message? How am I going to outreach to them? Am I going to message them? Am I going to text them? Am I going to send them a LinkedIn message? What's going to be in that? How's it going to resonate? Like, it's the identical process. You With the goal of maybe building a network as opposed to selling a product so

SALVARAAJU: Yeah, yeah, spot on, Tim. Yeah, look, in fact, it's an exact replica of all the outbound activities people do on the sales. You can actually use the exact tools they actually use. Okay, for prospecting, you can actually use the prospecting tools. There's a lot of tools out there. If you want to get your email editing, I can go to the tool. Put your name, put your company, and get your email. Then I can actually email it to you, and there are good email delivery platforms that are free to use. You can actually integrate with Gmail, and then you send them, and you make sure that email gets delivered.

TIM: Yes.

SALVARAAJU: And you can craft the right tone, write content using LLM, and then you can actually put your projects together and then try to get a coffee. I think it's the best way to actually go after the right kind of employer, especially the team that you want to work for. It's a very good strategy to actually execute; again, it just takes time and patience, but it's worth going that path, yeah, because that path is very likely taken by few. so you'll get noticed

TIM: Exactly, it's not a channel that's already been decimated, so I wouldn't hammer people's LinkedIn with your CV and say, Please give me a job, because every hiring manager will hate that, but a customized, crafted email that's personalized to them that somehow refers to something they've done and adds value to their life and says, Hey, a coffee? Would you be up for that sometime? in a very polite, friendly way that is going to land because almost nobody's doing that.

SALVARAAJU: Yeah, yeah, that's a Yeah

TIM: I'm interested in one final thing, actually silver, so if you had a magic wand, which AI almost feels like magic these days, so let's say AI or anything else you've got a magic wand, how would you personally fix the hiring process?

SALVARAAJU: Look, if you need to try to map how the hiring process actually works, just before I used that magic wand, usually there is a requirement in the company to execute something, so they come up with this: Okay, so we need to hire someone. So somebody writes jobs and roles and responsibilities, and they give them to the HR department, and the HR department then sits with the manager and then Hey, what are the prescreening requirements I have to do, and then they get all their requirements? What are the criteria? The applicants must fill in, and they go and advertise because it's mandatory to advertise in the job in the portal by law, and they started getting applicants through a lot of these different channels right linked. In emails, there are a lot of other ways you can actually get a CV from recruitment agencies. And then they have to have a prescreening, and after the prescreening, then they have a different set of rounds; for example, they have a technical evaluation or task assessment, then they have a cultural fit round, and then it goes through at least six to seven rounds before they get actually hired. It's a long process. It just takes time; a lot of effort went into it, and it is actually also costing the employer as well. Okay, the magic one I want to have is, okay, how do I make sure I can actually hire a candidate in just two weeks time? Okay, I want to have that agility, and if AI or any piece of technology would come in to streamline this long, lengthy process to make it efficient, at the same time I want to make sure I get the best candidate. that is culturally fit and also technically sound, okay, soft skills Yeah, and then again, the criteria depend on different organizations, but if you can actually shorten the recruitment process from months to a few weeks or a few days, I want to hire somebody in 10 days. Can you give me the best data? That would be the magic one I want to have that increase the organization agility given that as a business we are always the business always at stake; there's competition; there's new technology being come up; the customer demands are actually shaping, so organizations need agility. Okay. They also need to have elastic labor demand so they can actually use the people that they need, and then they actually go and build things for the customer so they can actually grow as a business, so that would be the ideal case. Thanks, and to provide a value, the HR process is the first start, correct? Before you get into the company, who's the first department in the organization you're interacting with? The HR team. Obviously, you want to make sure that team is extremely efficient and effective, and then if you could give some kind of a magic wand where they could be efficient and effective, go for it.

TIM: That's a great vision of the future, and personally I don't think we're that far away from it. I think hiring will be that simple within the next couple of years based on the rate of change of AI and how It's already at a level that it could solve, I think, a lot of these issues with hiring. We just need to implement the software and the application layer to do it, but I don't think we're that far off personally.

SALVARAAJU: Yeah, look, that's extremely positive. I think that if the things are happening, I know that the market is actually shaping up, which is a good thing. Hopefully it will change for the better.

TIM: 100 okay, Silva It's been a really fascinating chat today with you, but really interesting to hear your insights across a broad range of different topics. So thank you so much for joining us on the Alooba Objective Hiring Show.

SALVARAAJU: Yeah, all right, thank you, thank you, Tim, thanks for having me, and it's good; it's great to be with you and then discussing a lot of interesting things. Thanks so much.