Rebuilding Engineering Hiring for the Rise of AI with Jonathan Durnford-Smith, Synthesia

Hiring is changing quickly as AI tools reshape how teams build products and how companies evaluate talent.

Jonathan Durnford-Smith, Head of Talent at Synthesia, joins Shannon to discuss why recruiting teams must become more agile and strategic. He shares lessons from scaling talent functions at high-growth companies and explains how AI tooling is shifting what great engineers look like today. Instead of focusing only on coding execution, his team now prioritizes product sense, customer empathy, and strong problem definition.

Jonathan explains how Synthesia redesigned its engineering interview process, resulting in a hiring approach that adapts more quickly while maintaining a high bar for talent.

Key takeaways:

1. Redefine engineering excellence: Evaluate engineers on product thinking and customer empathy, not only coding execution.
2. Encourage AI collaboration: Allow candidates to use AI coding tools during interviews to better reflect real work environments.
3. Define roles before hiring: Spend time clarifying the real problem the role must solve before sourcing candidates.
4. Challenge hiring assumptions: Talent teams should question role definitions and push back when something feels misaligned.

Timestamps: 

(00:00) Introduction

(00:48) Meet Jonathan Durnford-Smith

(01:48) Why agility matters in recruiting during the AI era

(05:03) How AI is changing engineering hiring

(06:00) Why product sense matters more than coding execution

(07:24) Redesigning engineering interviews around real business problems

(09:14) Why engineers must be comfortable using AI tools

(11:36) The signal great engineers show early in interviews

(13:52) The culture principle that enables agility

(16:30) Aligning hiring with company strategy and leadership bets

(17:29) Why recruiters must push back on hiring decisions

(20:25) Why hiring processes should be revisited every six months

(24:29) Where to connect with Jonathan

Transcript

Jonathan Durnford-Smith (00:00):

Talent's not a support function. It's like a growth lever. It's a strategic function. So I think I definitely went through the first few years of my career being a bit of a people pleaser, kind of like, "Yeah, what do I need to hire? What do I need to execute on? " And I think maybe as I've matured in my career, the realization is that recruiting isn't just filling jobs, it's building capabilities.

Shannon Ogborn (00:22):

Welcome to Offer Accepted, the podcast that elevates your recruiting game. I'm your host, Shannon Ogborn. Join us for conversations with talent leaders, executives, and more to uncover the secrets to building and leading successful talent acquisition teams, gain valuable insights and actionable advice from analyzing cutting edge metrics to confidently claiming your seat at the table. Let's get started. Hello and welcome to another episode of Offer Accepted. I'm Shannon Ogborn, your host, and this episode is brought to you by Ashby, the all-in-one recruiting platform, empowering ambitious teams from Seed to IPO and beyond.

(00:58):

I am super excited to be here today with Jonathan Durnford-Smith. He is the head of talent at Synthesia, based in London, and he's someone who has spent actually a good portion of his career inside some very high growth environments. Before Synthesia, he led engineering product and applied hiring at Google DeepMind during a really major period of growth.

(01:20):

Earlier in his career, he was employee number 37 and founding recruiter at QuantumBlack, the AI consultancy that was later acquired by McKinsey, where he helped scale the company from 37 to more than 350 people. So it's safe to say that John has seen a thing or two when it comes to building talent functions while companies are moving very, very fast, which is kind of the era that we are in now, which we'll get to in a little bit. So John, thank you so much for joining us today.

Jonathan Durnford-Smith (01:45):

Thanks very much for having me, Shannon. Yeah, really pleased to be here.

Shannon Ogborn (01:49):

Well, like I was mentioning just now, we are going to be talking today about circumstances where companies are trying to move very, very fast, especially in this. I hate using the cliche like age of AI, but it is the reality of what we're in. And we're going to be talking today about agility and why that matters so much to talent teams and hiring teams right now. Before we get into how this has resonated for you all, would love to take a step back and hear a little bit more about why you feel like agility matters so much in hiring today.

Jonathan Durnford-Smith (02:24):

Yeah, of course. So I think the landscape of hiring is changing dramatically. I think a lot of that is, as you say, powered by AI tooling and a lot of these changes in the market for what recruiters can use essentially to improve on productivity and other things. What it's also meaning is actually the very essence of a role is changing across entire organizations. So recruiters now have to be able to pivot very, very quickly and learn even more quickly, I think, than in previous years, to be able to very quickly understand what is it that we need for the future. And it's very hard to do that because if you don't have a crystal ball to be able to predict the future, then a lot of it is testing really, really quickly, iterating really, really quickly. And I think the people that can do that and nail it will be the ones that kind of can get to the best talent the quickest.

(03:15):

And there's an awful lot of pressure to do it. I think a lot of talent teams right now have a lot of pressure to grow really, really quickly, but to keep that bar high. I could ramble on about that topic for hours, but I'll save that for a later question.

Shannon Ogborn (03:27):

I think one of the cool things about this moment though, specifically for talent teams and recruiters is that this is such a great opportunity for us. I still put myself in there. I'm still on us to really position ourselves as leaders at a company, as thought leaders, as people who are important to moving the business forward. So I actually think a pretty big opportunity presents itself when it comes to showcasing our agility.

Jonathan Durnford-Smith (03:57):

Yeah, I totally agree. And I think the times where superstructured organizations that have almost like by design a service recruitment team is going to die completely. I don't think that pivots quick enough, and I actually don't think that's going to be the way of doing things in the future by any means. And I actually find that quite inhibiting. And a lot of recruiters should feel and hopefully do feel more empowered that they are like a strategic edge for an organization now beyond justice service. We throw you candidates from over the fence, you throw us feedback in the other directions. So it is definitely the sum of a company's parts will make it successful, and that's largely the people that you bring in. And I think more recruiters need to have that realization, we're not a service function, we're largely a partner to a business and a strategic function.

Shannon Ogborn (04:51):

For sure. I think we have to believe at first before we can convince others to believe it. So I think that's a little bit of the state some companies are in, some have fully progressed, and I think the rest will come along for the ride. I know that you all have done some interesting things when it comes to agility or have some interesting philosophies there. How has that resonated with you all?

Jonathan Durnford-Smith (05:14):

If I could put out a specific example is around engineering hiring. So we need to hire pretty aggressively in the next couple of years. We're around 650 people right now. Across the entire organization, we need to hire about 300, 350 over the course of the next 12, 28 months. So pretty aggressive. 120 of those is engineering. Engineering is a role that's arguably the most affected by some of this change in terms of AI tooling. So you've got things like Claude or Cursor or other generative coding tools. That means that ultimately what we want an engineer to do now is quite different to what it was previously. And by that, I mean previously we would look for in a coding style interview execution, so how clean is their code? How well do they know their syntax? How well do they know data structures, this kind of stuff.

(06:03):

It's actually increasingly becoming more of a, we don't care so much about code execution anymore because there are tools that will do that for you. And it's becoming more, have you got customer empathy? What is your product sense? We talk about product sense a lot here, and we're still trying to get to the root cause of what the true definition of that is. But I break it down as customer empathy. Are we building things that actually get to the root cause of a customer problem? Or are we just saying, this is kind of a cool idea and a cool solution, let's build it. And actually we're a business at the end of the day. So I think a bare minimum for any kind of software business, you listen to your customers and you build what they want. That's kind of the bare minimum. And we are looking for engineers that can kind of have that mindset.

(06:46):

They're almost a little bit, a lot of people say forward deployed engineers or product engineers. I'm not a massive fan of just re-badging things for the sake of it, but I just think they're still software engineers. They just need to have a slightly different mindset now than they used to. So we've completely changed the way that we hire in engineering and we're continuing to iterate really, really quickly. And some might point out that, oh, actually it needs to be consistent and it needs to be repeatable for fairness. Yes, but if you make small tweaks here and there, an A/B test and feel like you're not putting anybody at a severe disadvantage, that's okay. You need to do that kind of thing.

Shannon Ogborn (07:24):

Yeah. What has actually gone into the changes? I know that you had mentioned some of the things that you were now looking for, but I'm curious how the agility mindset has really applied. And if you were a talent team or you were advising another talent team on how to do this, what should they be thinking about?

Jonathan Durnford-Smith (07:47):

Yeah, we started with, it was actually, I noticed our SVP of engineering and our CTO and others were talking more and more around Product Sense and this thing that kept coming up in interviews. And I kind of double-clicked on it and was like, hold on a minute, we're saying this term a lot now and we've not really defined what it is. And so we should probably do that. And it was just a case of getting a group in the room with a couple of my team as well for us to really dig into actually what are areas in the business right now where we are failing? What are areas where we are moving slowly? What areas where we could be doing better in engineering? And using that input to mold, okay, if we could hire 10 perfect people to fix those problems tomorrow, what would the attributes be and how would we hire for those?

(08:28):

And really doubling down. And I'm going to sound like a bit of a dinosaur right now, but if you take all of the tooling to one side and everything you hear about all the cool new ways of recruiting, everything still boils down to defining the role and what's needed. That's still going to be a consistent thing. And I think that's maybe the human element that perhaps AI just can't replicate or can't necessarily get to the root cause. Maybe it will one day and we'll all be out of the job, who knows? But I do think defining something and getting it really clear before just jumping in and talking to people and by people I mean candidates is kind of like 80% of the problem I think in my experience when it comes to recruiting. So we did that. What came out of that was essentially, well, we need our engineers in- house to be much more comfortable with AI tooling.

(09:14):

Our ability now to very quickly prototype a feature should be in a couple of days, it shouldn't be in a couple of months. And we're actually spending probably more time requirements gathering and defining the problem that we're working on now than actually building what it is we believe the solution will be. And that's a totally different way of doing it. No longer do you have five engineers building a feature for three months. We've actually had cases where one engineer can do that now in a week or maybe two weeks with context and really, really good requirements gathering. So we've taken that, we've put it more into our interview process. The first one is we still do a coding interview, but it's all integrated with AI tooling. So if somebody wants to use a code generation tool, then they can use that. That's totally fine. And what's been really interesting is actually we've had several candidates join those calls and say, "Oh, can I use Claude or Copilot or whatever it is that they want to use?" And we say yes, and they're really shocked.

(10:07):

And what's really funny is some of the candidates then are immediately more nervous because they worry that that's like a catch-out. It's going to be like a gotcha moment where they've started using that tooling and we're going, "Ah, you've actually cheated, sorry, you failed." And that's totally not the case, but hearing that feedback was really interesting because I was like, "Is this the state of tech recruiting where we're at right now where candidates are thinking they're going to get caught out at every point?" Do you know what I mean? I

Shannon Ogborn (10:30):

Just think people feel like they're set up to fail generally with interviewing. And as an organization, you have to do a lot to back people into the idea that even if that feels like the case elsewhere, that's actually not the case here. And there are plenty of companies who do have kind of gotcha moments. And I think it is kind of necessary to reiterate this is not a gotcha. This is what we're doing internally. And I think one of the important things that you said is that people internally are using AI tools. If you expect candidates to be well versed in AI tools in whatever capacity, whatever role it's in, your company better be doing something around that. You can't just expect everyone to be doing it and your company's not doing it at all.

Jonathan Durnford-Smith (11:18):

Definitely. Yeah. Yeah. And I think that's a really good point is I think some companies will be guilty of hiring almost in a way where you're hiring for a profile that doesn't even fit in your org right now, as in you're not practicing what you preach and that's not great. So I think that's a really good catch-up. And I suppose to build on that, what we're then looking for really is people may ask, well, if you're not looking at quality of code or how they're coding or what their syntaxnology is like, what are you looking for? And our focus has largely been around, yeah, maybe they built some codes, but can they explain what they've built? What's the reasoning behind what they've done? Are they actually answering the problem? And at the very beginning part of that interview, most of it for us should be requirements gathering.

(12:01):

And so that should just be the candidate asking lots of clarifying questions to the interviewer, kind of like, "What are the user expectations? What is the timeframe that we have here?" These kinds of things. And really that for us is a big signal even at the very beginning of that interview, have they just accepted what we've told them and they just go and execute blindly or are they questioning it? Are they challenging it? And actually the stronger candidates will say, "Actually, I think the brief has missed this, or actually I think maybe we should take this route." And that's really interesting. People challenging thinking is a big thing for us and that's slightly baked into our culture as well. And the other important thing was that we needed to be careful with how we frame this internally. So some of these things won't work and that's fine.

(12:43):

It's not going to work first time. As long as our leadership are on board with the fact that, look, we're going to experiment with things, we're going to iterate very, very quickly. Some stuff's not going to work. Some things will screw up. Some things will work really well. Some things will be like light bulb moments where we're like, "This was amazing." But if you set up that expectations and say, "Look, we need your buy-in, we need to calibrate, we will need some time from engineers in- house. We will just need to quite rapidly iterate on things." That's totally fine. I've not had any pushback. And for me, that's almost like the TA team mimicking the way that we should be working. A really effective TA team should work in a style that compliments the team that they're working with and they're partnering with. So that's actually worked really well is kind of, I think certainly in engineering, they quite like the way we work now because they're thinking, "Oh, you're just thinking through problems, you're iterating really quickly, you're putting in something new, you're trying it, " and that's resonated really well.

Shannon Ogborn (13:35):

That's awesome because part of being agile is teams working together in partnership. You can't have one team, super speed agility and the other team moving at a snail's pace. That will catch up to you in a bad way. But you just mentioned something about culture, which based on a lot of the things you're saying, it seems like a lot of culture elements are behind the success of this. How does culture impact what you've built and how you're operating?

Jonathan Durnford-Smith (14:12):

Yeah, it's a great question. The word culture gets thrown around a lot, doesn't it? And I don't think people spend enough time thinking about what does it feasibly mean? For us, one of the big things I noticed on joining the company was leadership and the co-founders, they push one of our values, which is own it and go direct, which is essentially like we don't want loads of blockers in the business. We want anybody at any level to feel that they have the autonomy to test and try and have an opinion on something. If you don't have a strong opinion, we'd question, okay, well, why not? What is it that you want to do? Where is the impact? And I think that's served us really, really well because it's allowed that agility. Now, I won't lie and say it hasn't stretched us because when you get to over a few hundred people, everybody running around like a headless chicken and trying new things can be hard.

(15:06):

And so there is a small element of structure that creeps in. But by structure, I kind of just mean before you go ahead and implement something or do something, show that you've tried and tested and show that there is potential impact, get whatever people, a small group, doesn't need to be hundreds of people bought in and then move forward and try it and do it. But beyond that, I don't think there needs to be a huge amount of restrictive structure. And that's kind of the balance that we're trying to find here is when I joined, it's probably fair to say it was around 300 people, Synthesia and we were still operating like it was like 50 people. It was kind of chaotic.

Shannon Ogborn (15:43):

Even our VP of marketing is like, we just at a minimum need to make sure we are all rowing in the same direction because if you're rowing this way and I'm rowing that way and someone else is rowing front and the other person's in reverse, then we're actually just going to stay in the same spot for months and months and months. So you at least have to have some directional trajectory, at least we're doing this all on a mission to X or we work this way principally, or at least you know what direction you're going in so people know, yeah, we have a lot of autonomy and we can move fast, we're not going to be blocked.

Jonathan Durnford-Smith (16:23):

Definitely. But

Shannon Ogborn (16:24):

We're not just doing things for no reason. We know what the reason is and we're sort of moving in that direction.

Jonathan Durnford-Smith (16:32):

Definitely. And a big unlock for us probably in the last couple of years is we've added to our leadership team with new CTO, new CRO, new CFO. And what that has brought is that direction. So I think you've hit the nail on the head there is that's the pivotal point is kind of these are the three things we want to do with our product this year, or these are the biggest bets that we want to make. This is the direction we need to head in. As long as in some way you are contributing to those, doesn't really matter. See what you can add, see what impact you can make, and that runs through the entire organization. So if I translate to what that means for TA, for example, it would be like, I have regular meetings with leadership on, within our bets, are we really hiring the right profiles that we need right now?

(17:13):

What is the profile that we need? And that changes quite rapidly, which can be like, I'm going to bang my head against brickwall, but it could also gives us in talent the right to say, "Do you know what? This doesn't feel right. This profile isn't right. This isn't what we should be looking for. These are maybe the areas that we should be probing into in the market or the profiles that we're looking for. " And I think that's another part of our culture is it's nobody has massive egos. So actually everyone kind of invites disagreement, obviously in a respectful way, but it's like if everyone just sat there and agreed, something's not right. It's like, hold on a minute, and we should be challenging each other's ideas. And I do think an ability to have an honest conversation with one another and say, "Do you know what?

(17:57):

Actually, I think we could do this better, and here's how I think we can do it better." And actually people not being super defensive with that, as long as, again, it's phrased in a respectful way and you're not saying, "Oh, your idea is garbage," which we never would. I think that's a big thing for our culture is the ability to not take things personally and everyone's kind of going in the right direction and it's fine to disagree. And I can sit there in a meeting with my CTO and actually tell him, "Do you know what? I disagree with what you're saying right now." And he's totally fine with that and vice versa. And I think that's really, really important. And I think sometimes in TA, I've certainly seen this in my career and have probably been guilty of it sometimes where you can go into autopilot and you're like, "I just need to fill these roles.

(18:36):

This is the direction that we literally need to go into. Let's just get there and just keep trandling along." But I don't think that's the right way of doing it anymore. I think you should question things. I think you should challenge things. And I think if it doesn't like, "Oh, what did one of our leadership say the other day? If it doesn't smell right, tell someone." Do you know what I mean? And I think everyone should feel empowered to do that. And especially TA now, like I said, this kind of links to the ... It's not a service function, it's a partnership and we're just as much able to call things out as others. Yeah.

Shannon Ogborn (19:09):

Not pushing back becomes a problem because then you don't hire the right person and even one wrong hire can set a fast growing company team back pretty significantly. So it's really important to give the pushback, of course, respectfully, because that's ultimately going to get, even if it takes a little bit longer or a lot longer, you're still going to end up in a better place yet because you're hiring the right person. So it's really, it's an important outcome. Obviously we want to hire the right person for the right role at the right time because I think people don't push back for a few reasons. One is fear,

(19:48):

That's a big one. And the other is because they're not motivated to think constructively or problem solving. It's like, that's where the service oriented piece comes in. Like you're saying, we're just going to fill the roles, we're going to move on, but that doesn't build generational companies. And if your company wants to be a generational company, you have to operate a little bit differently. So got to hire the right person. But I'm curious, after all of this and after the experiments you've been doing, obviously you probably don't have a lot of hard metrics yet, but what are you seeing directionally and how are you moving forward from there?

Jonathan Durnford-Smith (20:31):

We've not had a bad piece of feedback yet from candidates in that we've been doing this for a good few months now. So we've definitely had a couple of hundred candidates go through the process and every single thing that I've heard from candidates has been positive in terms of this experience. And by that, I think that's largely because we are practicing what we preach. We're really, really clear with our prep to candidates, we're really clear with our expectations. And I think candidate experience is kind of paramount in a market as things are right now where things are so competitive, you can guarantee that any good candidate is talking to five other companies and they will probably have at least a couple of other offers. And the one that stands out will be, yes, maybe the one that's thrown a ton of money at them, but also maybe it's the one that they've actually said, "Do you know what?

(21:23):

I've actually really enjoyed that process." And we had a couple of candidates that have referred other people that they've worked with before they've even started with us, for example, and they're mentioning us to other people. We've had a candidate that didn't join us, tell their good friend to interview with us regardless. They actually told us that. So that was amazing as well. And then another big positive outside of just results of candidates was it surfaces a lot of questions. It forces us to have some of these difficult conversations faster. Are we doing the right thing? Is this right? Should our expectations be higher here? Where is our bar for talent and all these different things? And that feels like an ever-moving part for everyone right now. But I think it's really helped us to dig into what are like ... If there were three attributes that you would want to hire for right now, what would those three things be?

(22:14):

And it's forced us to calibrate and forced us to get agreement from all of our hiring leads, all of our interviewers on this is what we expect. And that's been really, really helpful, just having that constant conversation. I think you have to be careful with how often you do this because it can bring in fatigue, hiring fatigue, decision-making fatigue, this kind of stuff. But I would strongly suggest that companies should be questioning the way they hire every six months from this point onwards. It shouldn't be any less than that because things will change so quickly in the next couple of years, not just in tech, I think in lots of different roles. So it's going to be a really, really interesting time. But yeah, I would really push people to think in that way.

Shannon Ogborn (22:55):

Yeah. The fatigue gets very real. There's a point where you move too fast too often, and then people are like, "I'm sick of this. I can't keep adjusting because then the recalibration takes so long that you're not even getting the right data." So any last thoughts though on this before we move to our more fun last questions?

Jonathan Durnford-Smith (23:20):

No, it's when you said fatigue there, I was like, yeah, I'm feeling bit of fatigue now actually. But yeah, with all the decision making and all the change. So I think that's a good point is that it's worth noting that and making sure, try not to break the entire group of people that you're working with at the same time. And yeah, I would just encourage recruiting teams to get into that mindset of constantly questioning, are they doing the right thing at different times? Is this the right outcome? And yeah, I think it's tiring, but the net positives are definitely there. And overstandardizing things and keeping things overly standardized for a long period of time is only going to make you loop. You'll fall behind, I think, to competitors very, very quickly, as I say. So yeah.

Shannon Ogborn (24:06):

It's all a great balancing act, isn't it?

Jonathan Durnford-Smith (24:08):

Yeah, it's not fun in recruiting, isn't it?

Shannon Ogborn (24:11):

Right. Well, we are going to move on to our last three questions. For those who are newer here, we have moved these to our extended version on YouTube. So if you are listening on Spotify or Apple, head there to hear John's take on Hiring Excellence, hot take and early career self. Well, we are coming up on our time. Where should people go to learn more about you and your work?

Jonathan Durnford-Smith (24:36):

Keep an eye on my LinkedIn. I'm probably not as regular a poster as I could be, but we'll regularly post thoughts and things on there every now and then. Yeah, feel free to drop me a connection. And I'm always keen to talk to other people in recruiting that have really interested ideas or things that subvert the norm. So yeah, by always reach out.

Shannon Ogborn (24:55):

I'm always hoping that listeners take guests up on this because everyone wants to connect, but then when someone offers, nobody connects. I'm like, please connect because they're so amazing. But I really appreciate you joining us. I think this is going to be really interesting for people who are trying to evolve at the speed that all the external tools are evolving, and there's obviously a lot that goes into that that you all are at the forefront of. So I appreciate you chatting with us about that today.

Jonathan Durnford-Smith (25:25):

Thank you very much. Yeah, it was my pleasure. Really enjoyed it.

Shannon Ogborn (25:28):

Thank you. This episode was brought to you by Ashby. What an ATS should be, a scalable all-in-one tool that combines powerful analytics with your ATS scheduling, sourcing, and CRM. To never miss an episode, subscribe to our newsletter at www.ashbyhq.com/podcast. Thank you for listening, and we'll see you next time.