A peek into Data Science at N26 with Gráinne McKnight
In this interview, I sat down with Gráinne, a Data Scientist at N26, to explore a bit more about her field, and how the data team works. Data Science hasn’t played a strong part in my history in tech so was fascinated to hear some of results. Read on for more insights.
Patrick: Hi Gráinne! Thanks for taking the time out of your day. Would you please introduce yourself?
Gráinne: My name is Gráinne, and with my accent, you can probably tell that I’m Irish. I’ve always loved mathematics. I studied my Bachelor’s in Mathematics, and then specialised in statistics. I started working in more traditional, conservative banking as a credit risk modeller. I loved the job. I got to learn a lot about economics, topics we didn’t cover in school. I decided I wanted a change, and came to N26 where I’ve been here for more than a year.
Patrick: How did you get into Data Science? At what point did you decide to move in this direction?
Gráinne: For me, data science is statistics. But maybe with more of an emphasis on the infrastructure and tooling built around it. Before I focused on specifying models with other people building the infrastructure. Now I’m responsible more end-to-end. I was nervous before making the change. However I’ve always enjoyed programming and learning about today’s modern tooling.
Patrick: What tools have you used?
Gráinne: I previously worked with tools like R, or SAS. At N26, we use a lot of python in the data team. This was my first time using many of these tools, although I did practice a bit before I joined. I feel, like most things, once you’ve been exposed to the tooling, you know which questions to ask. Knowing the right questions means you can find answers online or ask the team.
Patrick: You’re right. We are so lucky to live in a time of such open information. If you can’t find it, you can post something on Stackoverflow. Or you write an article, contributing back to that body of knowledge.
Patrick: Why don’t you talk about a Data Science project that you’re excited about here are N26.
Gráinne: I’m currently working on a few projects. One that I’m particularly excited about is modelling financial crime risk. My background in credit risk modelling has helped. I’ve learned a lot about how people attempt to defraud each other. It’s fascinating but also slightly depressing. We are lucky have so much freedom to innovate on how we detect financial crime. We have very modern tooling, enabling us to quickly build and iterate on our complex models. Topics like fraud are a constant ongoing red-queen game. Our tooling enables us to move quicker as the environment and behaviour changes.
“We have very modern tooling, enabling us to quickly build and iterate on our complex models.”
Patrick: Great. What don’t you share a bit more insight into how the data department works? What roles do you have and what do they focus on?
Gráinne: The data team is, unsurprisingly, focused on data. We have three broad groups — Data Engineering, Data Analytics and Data Scientists. I love how we work because we are all closely related and we work very collaboratively. This means we learn from each other and support each other. Data Analysts focus on learning and interpreting data for business projections. One example might be understanding future projections of growth. Data Scientists focus on building models to learn from the data, rather than analysing it. For example, we might develop a model that can learn how to recognise suspicious account behaviour. Data Engineers work at optimising our data pipeline. An example might be how data arrives, or how to feed data back into the system.
“I love how we work because we are all closely related and we work very collaboratively.”
Patrick: That sounds very interesting. What’s an example of how you work together?
Gráinne: As an example, when we start an initiative, we’d work very closely with Data Analysts to understand where the data is and what is relevant. I like to think of them as librarians of the data warehouse. We then iterate and test our models. As a model becomes closer to product ready, Data Engineers set up the right pipelines and then we integrate these models into the product.
Patrick: It sounds like a very fluid collaboration style. I’ve seen companies where those groups don’t at all interact. In your team, it feels like you get the right people in different roles work together at different points very organically.
Gráinne: Exactly.
Patrick: What do you think is exciting about the field of Data Science today?
Gráinne: I get really excited when I personally learn from Data or when Data shows an insight I never had before. One counter example might be our current chatbot. Our chatbot has been very successful at answering many queries. It can help people connect their questions to existing bodies of knowledge. However because English is very hard for a chatbot to truly understand. I can say, in a way, I am much smarter than the chatbot, though much slower at responding!
Gráinne: An example might be when understanding credit or financial crime risk. Data Science teaches me more about what is risky behaviour and what is not. A good analogy is seeing stereograms. You can see patterns but it’s hard to understand what the meaning of those patterns are. Data Science teaches us to recognise those patterns. It also helps us understand if those patterns are positive or something to avoid. It’s like once you see it, you can’t unsee it (laugh). That’s what I find very exciting about Data Science. I’m also excited about the potential of Data Science at N26. We’ve only started to scratch the surface of what smart banking could be like. We have so many ideas of where it could go. We need more time and people to work on those ideas.
“It’s like once you see it, you can’t unsee it”
Patrick: I agree. We have a good basis to build upon and we have many more experiments to try. What are three things you’ve learned since joining N26?
Gráinne: The first and most difficult was learning how to make decisions. I would describe where I was previously working as very, “top down management”. In my previous employer, it felt like others made decisions for you, rather than by you. At N26 we have more end-to-end responsibility. We have more autonomy and decision making power. You might think for someone who builds decision models, this would be super easy but it’s not.
Patrick: Shifting to leverage autonomy, is indeed, sometimes hard. What are two other things you’ve learned?
Gráinne: My programming and infrastructure skills have improved a lot. I’ve learned a lot more about Data Engineering because I’ve most recently been working with a Data Engineer. This area blows my mind as there’s so much to learn!
Patrick: Do you have a book you could recommend?
Gráinne: There are so many books out there.
Patrick: Do you have a favourite?
Gráinne: The Elements of Statistical Learning, which is available as a pdf download from Stanford University’s website.
Patrick: I’ll add it to my reading list as I haven’t read it. Finally, what would be a piece of advice you would offer to someone interested in Data Science?
Gráinne: Maybe I’m biased because I come from a background of mathematics and statistics. It’s important people have a fundamental understanding of algorithms. Many people romanticise Data Science today. People often think all we do is draw on whiteboards, sketching algorithms. That’s only a small part of what we do. There is also data infrastructure, data pipeline construction, model performance monitoring and much more. I enjoy that, but you have to accept this will be a large part of your time as a Data Scientist.
Patrick: That sounds like a lot of work. Even though compute, data and network is cheaper and faster today, I can see how patience is still very important. Thank you very much for your insights into Data Science, your team and how you work at N26.
Gráinne: Thank you.
Interested in joining Gráinne at the bank the world loves to use?
Join us on the journey of building the mobile bank the world loves to use, and take a look at our open roles here.
