Nobody Has Done This Job

Roopesh: The role of the recruiter, really, in its current form — whether it's campus, professional, or exec — will never be the same. And this is still very early days.

Iris: We can hire people that are relatively earlier in their career, because AI is actually giving them the opportunity to become experts much, much faster than a few years ago.

Serena: Welcome to Deep Geeks. Today I'm joined by two people leaders who have first-hand experience in hypergrowth environments. Roopesh and Iris, welcome.

I'll open by getting to know you a little. My question to most CHROs, Iris: what made you say yes to this job? Bring us back to that moment.

Iris: In your mind it's tough. For me, I love it — and I genuinely mean this — it's my hobby. My work is my hobby. I really like solving tough problems, and I really like connecting to people on a personal level and getting to know them. I'm doing what I love.

Serena: There's a saying: if you love what you do, you don't work a day in your life.

Iris: It's absolutely true. Obviously there are days I feel the grind — WEKA is still operating as a startup, and I'm sure it's the same for Roopesh at Nscale. But it's fun. We always remember to add the fun factor to our day-to-day.

Serena: On that note, Roopesh — what about you? I've seen the tremendous growth at your company. Was it 100 to 950 in 14 months? Correct me if I'm wrong. Just shocking numbers.

Roopesh: Since that email was sent, we're at 1,050 now. It may increase during the course of this conversation, so I'll give you a live tracker.

Serena: What does your typical Tuesday morning even look like?

Roopesh: Each day is a bit of a blur. Like London doesn't have seasons anymore — the weeks don't have names.

It's never been busier. But to Iris's point, I do think there's a direct correlation, and this is what I tell my team, between enjoying the work you do, enjoying the people you work alongside, and productivity. Over the years, the ones who've performed best are the ones who come in with a skip in their step. There are good days and bad days, but generally speaking, those who enjoy it tend to be more productive.

My typical Tuesday is trying to solve for the incredible ambition Nscale has right now. They've raised a lot of capital — you've seen the Series B and Series C — and that can happen in months. But scaling a business takes months and sometimes years.

Scrappy is a good way of describing it, but it's an overused term, because scrappy can sometimes imply careless. This is not careless. Otherwise you wouldn't scale to these numbers and have the success we've had — we're only a two-year-old business. It's not careless, it's scrappy. It's doing what we need to do now to deliver what the business needs, and that evolves over time and becomes more sophisticated. But it's pretty relentless.

Serena: You mentioned good days and bad days. What does a recent good day look like? Give us a little behind the scenes.

Iris: For me at WEKA, a good day is a day where there's progress. Since I started at WEKA three years ago, I can't remember a week without significant progress, and that makes me tick — and my team as well. We're very much execution-driven. That's one of the competencies we measure the team on.

Roopesh: I love that, Iris. Good days are when progress is made. Sometimes progress is small nodes in certain areas, and sometimes it's colossal. But as long as you're moving in that direction — it happens in increments. That's a good day.

Iris: I'm curious, Roopesh: Nscale is a relatively newer company and you grew super fast. Can you tell me about process versus operating without a process? You need to grow ultra fast, and I'm sure your recruiting infrastructure isn't fully baked yet. How do you go about it?

Roopesh: I thought Serena was asking the questions! But that's a really good question.

To put this into perspective: January last year it was 100 people. When I joined in January this year, there were 300. We're now at 1,000.

The worst thing I can do is let process overpower innovation and curiosity. You only get to where you are if you have innovative thought leaders who are curious — curious about delving into the unknown, curious about doing things nobody else has done, breaking the mold.

So I came in saying: it can't be right that some people are having two interviews and some are having five, that some people are taking interview notes and some aren't, that some are doing testing and others aren't. Yes, you've grown really fast and you've been successful — but we need to add some rigor into the engine.

Number one is justifying why you're doing things, and communicating the outcomes of your change rather than the change itself. Putting a structured interview process in place seems obvious — everyone's nodding — but you're still changing the way very successful founders have been doing it, and if it wasn't for them, we wouldn't be where we are today. So getting sponsorship and showing the outcomes before you introduce the change is the theory of how I've been managing it.

Serena: I led people analytics in large organizations for a long time, so for a while I was the scientist testing what works in hiring. How many interviews do we need? I have a hypothesis — let's test it with data. We believe this question matters more than this other one, and then we let the data tell us.

Does data play a role in how you decide? You're growing so quickly it almost feels like you don't have time, but how do you actually know what works when you decide what to change, remove, or add?

Roopesh: Data fundamentally plays a part in our process. Part of this was justifying why a TA dashboard needed to be built, and why we need to identify and agree upon key metrics — not just the traditional talent acquisition metrics, but organizational health too.

For example, testing the hypothesis that you don't have to hire only from five target companies for people to be successful. The data shows how fast people are promoted, how they progress, what their performance looks like. Where most companies check in on performance yearly, we do it quarterly. We iterate quickly, and that gives us more data points to determine whether we're making the right decisions. We built our own dashboard to be unique to what we're doing.

Serena: On dashboards and organizational health — a lot of companies are starting to pay more attention with AI. How does our company look tomorrow? Iris, you work so much on culture. How do you define organizational health at WEKA?

Iris: We pulse the organization every year, sometimes more. But for the most part, you feel it. My team and our leaders are talking with their teams. We're building a good communication line between the business and HR. Our HR team and the recruiting team have a seat at the table, so we work very closely with the business and make sure we're constantly improving. We make mistakes too — but we know how to iterate fast and make changes.

There's a way to check organizational health through data, and that's our main pulse. But there are things you can't really measure. You need to talk with leaders. You need to talk with the team, with the ground — with the interns, the new grads, the rest of the employees, the people who aren't coming into a physical office and sometimes feel more isolated. We try to meet with them as often as we can and understand how they feel, how they experience WEKA.

One of the things we really encourage across our entire population is to bring their true self to work. That's one of the reasons I joined WEKA. We have representation in over 20 countries, many different cultures, locations, and time zones. We need to find ways to connect everyone.

We've built a lot of dashboards — recruiting dashboards, HR dashboards. We measure attrition, mobility, tenure, organizational structure, performance. We have a manager academy. Many, many different dashboards. But at the end of the day, that's not creating or measuring organizational health. Organizational health is data-driven, but also—

Serena: Conversational.

Iris: Conversational. What we feel when we speak with our population.

Serena: What's your favorite question to ask the interns?

Iris: I like telling them that what they think they'll do next won't necessarily be what they do in their career. So my main coaching point is to be open-minded, to focus on developing really good collaboration internally, to be curious, to ask many questions, to get to know as many people — but also as many functions — as possible.

The fact that you studied neuroscience, or computer engineering, or human resources, doesn't mean that's where you'll end up. Especially now, in the AI era. There are unlimited opportunities and possibilities for what an individual can do.

So: be open-minded and curious. That's my main guidance. And I always ask them what they want to do, because I'm curious to watch their growth and see how they evolve.

Serena: Couldn't agree more. Roopesh, what would you add?

Roopesh: It's difficult to add to that. Can I just say ditto? Everything Iris said, I agree with.

We're slightly earlier in stage — we've only ever done one pulse survey here, and the second is due to go out in the next few months. There's a lot of learning from that: not only whether people are willing to participate, but taking action on the responses we got.

From a culture standpoint, what's really apparent to us is this: it wasn't that long ago this was 50 people. And when it's 50 people, culture travels by osmosis. Everyone's sat in the same room. Everyone hears from the founders, hears their reasoning around why we're here, why we created this, and the journey we want to be on.

But when you get to 100 and 500 and 1,000, it has to travel with a level of purpose. When someone says, "I actually didn't know that was happening," or "I wasn't aware of that acquisition" — at that point, you have to be deliberate. It has to be repeated. You have to articulate it, not assume people understand.

From a leadership standpoint, we have to be visible. We have to communicate. Culture is people. It's not the website, it's not the PowerPoint — it's the people. And if you have one team of people who all come from the same company, guess what? That's the culture you have. So being mindful of that, making sure leaders are inspiring from the very top, keeping their finger on the pulse across their teams, and then feeding back — that's key. The nimbleness of who we are as a company has to be translated into the nimbleness of how we communicate.

Iris: I totally agree. The people are the culture. And culture is evolving. As you said, Roopesh, at 50 people the culture was completely different than what it is today at Nscale — and similarly at WEKA. Every six months I feel, and I'm sure others feel, that it's essentially a new company. We're evolving as a company and as a culture, in a good way. What fit the company culture three years ago doesn't necessarily fit today.

But at the core, since I joined, the core has stayed. We tend to hire leaders and managers who are doers, who roll their sleeves up, who aren't just leading the team but are very involved — who want to know what their teams are doing, who want to contribute and help and support. That's the core, along with the execution-driven approach we've always had. We have a culture of demanding execution, but we cover it with a lot of empathy and care to make it a little more fun day to day.

Serena: How do you evaluate for that? It sounds so difficult. Yes, empathy and care — who wouldn't love that? But especially when you're hiring the next set of senior leaders and you want them to embody those values, how do you know?

Iris: First, lots of conversations. My team isn't here to speak only with the executives or the VPs. We speak with and pulse as much of the organization as possible. We sample many employees on a regular cadence, and we listen.

I always say the values, the culture, the behavior, the collaboration, and the communication are 50% of performance.

Another thing that helps is the performance cycle we have — upward manager feedback, where all employees provide feedback to their managers. And the annual engagement survey, which has a section for managers, and we learn a lot through it. Luckily our managers are really appreciative of getting and giving feedback. They invite feedback for themselves, and they want to improve constantly.

Serena: What about you, Roopesh? As you think about the senior leaders you need, how do you actually evaluate that? And can you give some tips to someone who wants to be a senior leader in this industry going forward?

Roopesh: There's no exact science to it, because sometimes the interview persona is not the same as the real persona — especially at senior level. You can be whoever you want to be in that interview. We've all seen it.

We're making a conscious effort here. It hasn't always been the case, and I wouldn't say it's perfect today. But again, I'm testing a hypothesis: do you have to have done the job to get the job?

The way senior search has always worked — and I've been in search most of my career — is: of course you do. If you're looking for a CFO, you're going to source people who have held that role at companies you admire, with some industry overlap, and then you assess for fit.

We need to test that. And we are testing it. We all know fantastic people who would be incredibly successful in the environments we're in who have never stepped foot in a data center. We all know them. But the chances of them getting through the process are very slim.

Roopesh: Unless you're prepared to be open to strategic introductions. The moment you have a senior search, there's urgency to fill that seat, so you tend to go after what you know. And if you're a hiring manager who's been wildly successful over the years doing it the same way — who am I to tell them to change?

So you need to do it off the ball. Meaning: when they don't necessarily need somebody in that role. It could be tied to succession, or to building the overall talent landscape. Why aren't you curious enough to meet incredible people who do the same kind of role, maybe from a different industry sector, and get to know them? We do events and roundtables, and all of a sudden they're warming to people they never thought they'd warm to.

There's also this: if we aren't looking at the obvious profiles, this cannot be just about "have you done this role at a company we admire?" It can't be — because just because someone's been successful in one environment doesn't mean anything. A different company is like a different planet. How are you going to survive?

So we need to codify who we are, what our culture is, and what we genuinely believe in, and tie that into our assessment process. Make sure interviewers have a defined role and responsibility. Nail down the competencies we believe will be successful in any given role or department, and then assess against them.

And it's working, in the sense that a number of people who would otherwise have been an obvious hire haven't made it through the process — because on paper they should have gotten the job. I think that's a win. They're being selected out for reasons beyond having done the job before.

Serena: Yes, very much. One of the things we started doing was a monthly coffee with the senior leadership team, introducing them — virtually and sometimes in person in hubs — to people they wouldn't normally talk to, people they wouldn't normally have thought of for the role. And guess what? Some of them did turn into successful hires. But many, many months down the road, when something opened up. That works, because when you only search when you need someone, you're limited. The time pressure is real.

Roopesh: We're all guilty of it. We're all hiring managers, and the moment someone incredible leaves and I have to replace them — how prepared am I? The theory behind it is real, but in reality: if you needed a people analytics expert, how prepared are you to hire someone who hasn't done the job before but may have the skill set? We're all guilty of it. There's learning for us all.

Iris: Exactly. But even if you hire someone who's super skilled and has done it before, that can't predict how they'll operate when they join the company. So if you ask me, I'd much rather hire for potential.

It's more important to hire people who are likely to have the attributes the organization is looking for — the competencies we measure our employees on. We need people who are execution-driven, who know how to problem-solve, who have ownership and accountability, who are aligned with our values, and who can learn and be adaptable. Especially now, with AI.

That counts at least equally to professional experience. It's much harder to find people who really match the organizational culture and values than to find people who are a 100% fit on professional experience. I'd rather have the right balance between the two.

Our values were a long journey — about a year-long program where the entire population participated in choosing them. We landed on ABC²: accountable, brave, collaborative, and customer-centric. And we embedded it, to Roopesh's point, in the interview process. Every interviewer has a kit of questions they can ask — and they can adapt them — that show whether a candidate has the potential to embody our values.

Serena: In your industry, AI talent is so scarce. You can't find a ton of people with the perfect fit and the right experience — they just don't exist. I'm curious how AI is changing what you do. From everything you've shared, it sounds like you're using a very human process, which I love. How is AI changing it, if at all?

Iris: AI is simplifying the process. We have data, we have dashboards. We can evaluate the process constantly, and much more easily. We're also able to analyze later on whether we hired the right talent. We can pull the data much more easily for one-year churn, how many people stayed in the organization, and whether the process we ran was a good one. That's how AI really helps us.

Serena: Roopesh, what about you?

Roopesh: It's completely transformed the way we go about our end-to-end recruiting process — and this is just the very beginning.

Competitive market intelligence gathering is done in seconds. Candidate identification, sourcing, screening at volume. WEKA and Nscale don't suffer from a lack of applications. Our brand is propelling, it's being amplified. More and more people are hearing about who we are and what we do, and the byproduct is far more traffic than we've ever had.

The traditional, subjective way recruiting has always worked is that you make a visual assessment. And despite how sophisticated a company is in its product suite, most organizations still work that way. We've had to automate that. Screening at volume, market mapping, talent intelligence, scheduling — all automated.

It's that repetitive first mile. Automating it gets us to the stage where hiring decisions actually have to be made. The interview itself has to happen — the questions, the feel, the fit, the culture, the enthusiasm. If we can take away a lot of that repetitive first mile and really focus on the human element, that's helped us enormously.

Don't get me wrong, we've come a long way and there's still a long way to go. But it's made a massive shift. I'd go as far as to say the role of the recruiter in its current form — campus, professional, or exec — will never be the same. And this is still very early days. Even now, when I'm interviewing for people across my team, if you don't have a level of fluency around how this has changed your end-to-end recruiting process, you don't get through.

Serena: Of course, you're not the only ones using AI. The candidates are too. How are you handling the 10x, 100x, maybe 1,000x quantity of resumes coming in? Is that a huge problem right now?

Roopesh: It's huge. We have a tool that helps us identify applications that are unreal, through various data points. It could be how the application is structured. Let me give you a recent example: we had one applicant apply 100 times for different positions. Same person, same contact details, but multiple different personas in terms of what they do.

That's an obvious one, but the tool identifies signals like that and gives a rating on how likely an application is to be AI-generated.

And look — video interviews, you have to keep your video on, because we know what can happen. There are certain questions we would ask, and we guide our interviewers that if during the interview they suspect something's not quite right, there are trigger questions they can ask. But it's not obvious. It's a huge problem in our industry.

Iris: Today the ATS systems are actually flagging profiles that seem fake, so it's relatively easy to identify. But we try to rely — and I'm sure you do too, Roopesh — on employee referrals and on sourcing, with AI for sourcing. The majority of our hires come through our own sourcing and our employee referral program. Applicants are considered and hired at WEKA as well, but I don't remember a situation where we started an interview process with a fake profile. We definitely have many more applicants now, and some of them are fake and flagged early on.

Serena: This is one of the wildest markets I've personally seen. Because of AI — and because you're hiring in one of the toughest, most niche corners of AI infrastructure. Let's close with one word to describe what it feels like to hire in this market right now. Iris, then Roopesh.

Iris: For me, if I had to choose one word: exciting. Super exciting. I can see the opportunity ahead of us.

If I can expand — we're not necessarily hiring experts in every area. It's very niche, and you always need to balance. We need to hire a couple of experts, and people who have the potential to become experts within a year. That's what's really exciting: we can hire new grads now. We can hire people relatively earlier in their career, because AI is actually giving them the opportunity to become experts much, much faster than a few years ago.

Serena: That's so optimistic. I hope the new grads and college students listening hear that from our chief people officer. Thank you for giving us hope. Roopesh, your one word?

Roopesh: Very similar word — I was thinking of invigorating. Let me explain what I mean.

I'm very old, and I've been doing this a long, long time. But I thoroughly enjoy what I do, and that's what Iris said at the top — she clearly enjoys what she does. I feel I've got my mojo back. The market's turned on its head, and what was once obvious is not, and AI is transforming it.

I honestly didn't think I needed to learn anything anymore, that I could still perform relatively well. And the fact that I'm being pushed, that I need to stay ahead, that people are looking at me for the answers — or at least to surround myself with people who have those answers — wow. I feel like I'm at school again.

For many, they'd think, "This is not for me. My learning days are done." But I find it fantastic that we're reinventing the way things have been done for decades. That's pretty cool.

Serena: Invigorating and exciting. There you have it. Thank you both so much for joining Deep Geeks today and sharing your insights and experience. I really appreciate learning from both of you.

Iris: Thank you, Serena.

Roopesh: Thank you, Serena. Lovely to see you again.

Serena: Thanks for listening to Deep Geeks. A huge thank you to my guests today, Iris and Roopesh. If today's episode made you think differently about how AI gets built or powered, share it with someone who needs to hear it. Find Deep Geeks on Spotify, YouTube, or wherever you get your podcasts. Until next time.