Hey Alex, can you start us off with an introduction to yourself please?

Sure, my name is Alex Spanos and I’m the Lead Data Scientist at TrueLayer. I work in our Credit team, whose mission is to make it faster and easier for any company to create products powered by banking data insights of its potential customers or users. I’ve been working at TrueLayer for the past couple of years. Prior to that I worked at another startup called Black Swan Data and got into data science per se around five years ago with IBM. In terms of my academic background, I’d say it’s a classical applied science background focussed on mathematics and statistics.

Was Data Science always the plan?

When I started uni, I didn’t have a clue what I was going to be doing in my professional career and if it would evolve into an academic one or something else. I really liked statistics though. I figured that whatever I was to do, would have to involve numbers and data. So, I did a Masters in statistics at the University of Leeds and it was about that time when the whole data science explosion began. There were no Data Science degree courses at that time, there were mainly Statistics, Computational Statistics, and Computer Science options. So, I tried to squeeze in as many Data Science related modules as I could, and it just coincided with that big Jupyter Notebook/iPython Notebook explosion and that’s when I decided Data Science is going to be my professional orientation.

What made you decide to start your data science career in such a large business?

So to contextualise, before joining IBM (just a bit of context), I had started my full-time professional career in geophysics, believe it or not. There were many Data Science elements in what I did, but they weren’t formally recognised as “Data Science”. At some point, I wanted to break away and move into a formal Data Science role.  I started looking to make the change around 2014, and it’s fair to say I had a hard time! I think that it was one of the hardest things I’ve done in my life, transitioning to a formal Data Scientist role because back then this industry, if you can call it that, was a bit like the Wild West. I mean, most businesses and candidates were not quite sure what the requirements were, you saw these “Christmas gift list” job specs floating around and Data Scientists were kind of trophy acquisitions. It was pretty hard to get a job if you didn’t have ‘Data Scientist’ on your CV. So, I had a bunch of interviews with a bunch of companies and one of them was IBM and that was my lucky break. It’s a company to which I owe a lot to, in the sense that they gave me a chance to get my first formal Data Scientist role and work with amazing people.

What attracted you to Truelayer?

I think that FinTech is a hotbed of innovation here in the UK at the moment and a very exciting industry to be in. There’s a lot of competition and regulation has created a lot of opportunities for innovation. So, that was one of the main factors, but more specifically, I was looking for an opportunity to set up a Data Science functionality within an organisation from scratch. I was also interested in getting more experience on the engineering side, so I was really looking for deeper integration within an engineering team. I guess the lure of the unexpected as well. It was really the right time in my career to take a risk and move to something not so well established, but which promised a lot. TrueLayer offered all of that and more. I guess, they’re also the intangibles. Throughout the interview process, I just got a sense that my personality was aligned with the culture of the company and that just clicked.

What was the Data Science landscape looking like at Truelayer when you joined?

Interestingly, I joined as the 30th, I believe, employee and there were no Data Scientists working at TrueLayer before. I joined alongside another Data Scientist at the same time. Quite interestingly, although we were brought in as Data Scientists, there was not much opportunity to do actual “Data Science” at the time we joined. The reason was that we didn’t quite have the infrastructure in place to allow us to be able to work with data at the scale we desired. So, my career at TrueLayer has taken a couple of turns for the unexpected. For a while, I have been working partly as a Product Manager, which has been an amazing ride but am now in the process of reverting back to doing what I truly love. This is thanks to the fact that for the past months we’ve built out the data pipelines and the commercial/legal framework to enable us to work with richer data. We’re now releasing new cool data-powered products based on Machine Learning!

How has the role of a Data Scientist evolved to accommodate the fast-moving startup world?

I like this question. I kind of like the conceptual breakdown of Data Science evolution over the years into three waves, and that resonates a lot with my own experience. I like to think of Jupyter Notebooks as symbolic in this context. We had the first wave, the pre-hype, pre-notebook era and pre-Big-Data era where data analysis, operational research, and predictive modelling was happening in a very confined industry-bespoke way. The job got done, but the tools, methodologies, and maturity varied a lot depending on the environment.

Then you had the second wave, which I like to call the POC (proof-of-concept) or “shelfware” or Jupyter Notebook wave, that started when iPython Notebooks exploded onto the scene, and I believe that was in late 2011. The PyData stack exploded along with them. Along with greater data volumes businesses were seeing, everyone got super excited about you know, “deep actionable insights”. But in reality, I think few businesses managed to extract significant and tangible value. It was great for marketing, but I think the businesses lacked the expertise to put Data Science into value generation during that phase. But I think reality kind of hit hard at some point and generic data and Jupyter Notebook manipulation skills stopped being sufficient.

I think we’re now on to the third wave, the “full-stack Data Scientist”, where one needs to be able to make things work end-to-end. This means from prototyping, down to deploying into production, down to monitoring and down to integrating into a company’s engineering infrastructure. This is the era we’re in today. Unlike the situation five years ago, I think companies have realised that those basic Python and Notebook skills are a bit of a commodity nowadays and the actual end-to-end skillset and ability is what really matters – even at the cost of sacrificing deep narrow expertise.

What has been the secret to hiring well at TrueLayer?

I think number one is intellectual curiosity and a passion for machine learning is a prerequisite for hiring Data Scientists and Machine Learning Engineers and we definitely look for that when we’re hiring. Being in touch with the community and with the developments in machine learning and being really excited about all that I think is a crucial attitude. In terms of motivation, we look for people who want to solve problems and get things done, rather than people who want to use specific tools. On the skills side, we don’t overthink it. Again, generalist Machine Learning and modern Data Science skills are what we’re looking but we’re not too prescriptive on that either. Also, evidence of learning in parallel is very important for us, as it shows dedication to one’s passion.

I’ve been leading two separate Machine Learning Engineer hiring processes, one in the UK and one in Italy. They’ve been quite different to our experience. Hiring in the UK, this was earlier on in 2019, at a point when TrueLayer was perhaps not as well-known as it is today, wasn’t the easiest task in terms of the level of traction we were looking for. There’s obviously a lot of competition for experienced Machine Learning Engineers and Data Scientists at the moment. It was a tricky process and some tough decisions made, which you can never counterfactually assess. There is no shortage of passion and enthusiasm at the moment, but there is a bit of a shortage of people with real experience in deploying stuff into production. So that being said, one needs to keep a very open mind when hiring. This is what I discovered and, above all, it’s the aptitude and the willingness to learn that really tipped the scale when making some of these decisions. We need to be realistic; you think you know what you’re looking for, but hiring doesn’t really work that way because you need to really keep an open mind. Regarding the hiring process for Italy, I was very impressed by the profiles and levels of technical expertise of candidates – even at a more junior level.

What advice would you give someone who’s joining a startup for the first time?

Take ownership. That’s the best piece of advice I can give. I don’t think you can expect anyone to tell you what to do. You should go and do it and take the risk (a calculated risk). This can be quite a difficult transition for someone who hasn’t worked at a startup before and maybe coming from a more hierarchical type of company. But in reality, I think taking ownership and not expecting things to come written down is all startups are about, I suppose. Startups are not for everyone. Compared to when working at larger, more structured companies, they require a larger investment of headspace. It can take over a larger proportion of one’s life, in the sense of it is a make or break situation.

What advice would you give to founders that are looking to build data science teams?

From my experience at TrueLayer, what worked really well was joining the company alongside another Data Scientist, and our skills, I think, complemented each other well. So, starting out a new function with two people worked well for us I guess on more than one occasions. But I also think that founders need to optimise for constrained resources when starting out. Having generalist skills to start out with, in my opinion, is much more valuable than bringing in someone who’s an expert, for example in computer vision, generally speaking. As founders can definitely attest to, the type of work that a new Data Scientist joining the startup will be undertaking is very likely going to be varied. So, having various skills and not having an inflexible attitude is very important.

What is it that you like about working in startups?

Being able to work in a non-prescribed way, being able to take those risks, being able to make decisions and drive product development with autonomy. So, it’s what everyone says right? It is true that in larger companies, bureaucracy can stifle innovation. My experience at TrueLayer has definitely shown me that startups can be a very rewarding career opportunity for someone who wants to take that step, to take risks, to drive innovation and to drive product development, but that also comes with a large degree of risk. You need to be prepared that there is a chance that you might fail.

What is one thing that you wish you’d known at the start of your career?

Good question. Two things. Firstly, I think imposter syndrome in this space is a real and big thing and deserves more acknowledgement and management. Secondly, I learnt that doing, rather than reading or taking online courses is a more effective way of learning. These are the two things that I found out over the course of my career which I wish were more apparent in the beginning.

What is the best piece or best use of data science/machine learning that you’ve seen?

I’ll answer this in a sort of a different way. What I think is amazing with our scene at the moment is this multiverse of new, diverse and inclusive communities of amateur and professional science practitioners. I believe these communities have the power to make a real impact on the world – something our world really needs right now, whether we’re talking about climate or health. I wouldn’t like to name a specific example, but I think these inclusive new communities and this new kind of hotbed of focus groups around the world has been amazing.

What field do you think data science and machine learning will have the biggest impact on in the next decade?

Being optimistic, I think health is a great frontier and I think the medical breakthroughs and all these wearables and IoT devices will create very rich data sets that can help us move faster into customised and personalised precision medicine. I think also you have the medical imaging side which is rapidly advancing and it’s a really good application for computer vision/deep learning. On the pessimistic side though, I’m a little concerned that we’re unprepared to deal with the societal impact what we refer to as AI is effecting in this new decade. I’m slightly worried that face recognition and computer-generated video/voice/text could find some rather utopian applications in everyday life.

What is the best piece of career advice you’ve ever been given?

That’s a difficult one. But at some point, I was told that by failing to prepare, you’re preparing to fail. So, going into every situation prepared is something I try and do every day. If you attend a meeting, you should prepare for it or not attend at all – and that’s something I didn’t necessarily do at the beginning of my career. One of my managers pointed that out to me, and it’s advice I always try to heed – even though it is not always feasible.

What is one book that has changed your life?

That’s also a really difficult one! Every book we read changes our life a little bit, so I think the impact is cumulative. But I have to name two important books, relating more to my professional life, that had a substantive effect on me..

One was Fermat’s Enigma by Simon Singh, which came at a time when I was a student in mathematics and I was a little bit disillusioned with ‘Why am I studying mathematics? This is boring stuff being taught in a classroom’. But after reading this book, I really discovered its true magic and had a newfound purpose in my studies. So that I think played a significant role in my life.

Additionally, when I read Ray Kurzweil’s book The Singularity Is Near (although I’m not a singularitarian myself), it just blew my mind in terms of how technology can drastically and completely change our lives, even imminently! It’s written in, I think 2005, but the author makes a bunch of predictions that are very interesting and describes many cool and not so cool effects of technology. So yeah, I recommend it!

Despite the disruption caused by the pandemic, Truelayer is continuing to hire across multiple roles in London and Milan.

Check out their careers page: https://apply.workable.com/truelayer/