At Wolt, data science is a vital competence that enables us to use machine learning and statistics for building scalable algorithms that power our products.
Our data scientists are embedded into product and business teams. They work with areas where forecasting, predicting, and optimizing helps our teams to make reliable decisions, run efficient processes, and provide a better customer experience.
Today we’re talking to Juan to get insights into the world of data science at Wolt.
Juan, tell us more about yourself 👋
Originally I’m from Colombia, but I moved to Berlin 10 years ago for my Master's and Ph.D. studies in Mathematics.
Having developed my data science competencies and worked in several companies, I decided to join Wolt and bring my data science perspective to the Marketing Tech team which is responsible for our marketing data platform, automation tools, and models. In a nutshell, this team strives to build a scalable marketing data platform in order to automate campaigns and spend optimization across channels — that’s what I’ve been contributing to for my time so far at Wolt.
How did you end up joining Wolt?
At first, my journey was very research-oriented — during my PhD, the main area of research was geometric analysis. This area is a mix between differential geometry, topology, and functional analysis. Even though it is not something I apply in my current position, the math background does give me solid foundations, not only to understand the methods we use but also to build new implementations on top of them.
At some point, I became slightly tired of the academic world and decided to do something else. With that in mind, I began to re-learn statistics — the start was a bit from scratch, but catching up wasn’t that hard thanks to my exposure to math.
All in all, I managed to start my path as a data scientist. After working for several years at different companies, I formulated my next career goal: a data science position in a product company with opportunities to work in the marketing domain which provides very challenging and interesting problems to be solved using innovative statistical methods. 🔎 Wolt was a perfect match for me role-wise. That’s how I ended up being here.
Let’s talk about your projects here at Wolt — what keeps you inspired?
These days, we build statistical models to optimize user retention and user acquisition, which is impactful for Wolt’s growth.
My favorite project is related to our Media Mix Model. Speaking briefly, our model aims to determine the optimal way of spending the marketing budget to get the best possible outcome. At Wolt, we have massive marketing efforts, that’s why it’s so crucial to provide insights to frame future decisions on budget allocation and to identify the best media mix. I’m inspired to build the foundations of the model as it’ll have a great impact all across the organization.
Can you tell us about the data science team at Wolt?
Currently, we have our central data science team of 14 people, and we’re looking to grow the team size in the near future. At the moment, our data scientists are based in Finland, Germany, and Sweden. With our remote working model, we welcome people in Denmark and Estonia as well. 🌍
Each data scientist focuses on their own product or business area. Some of our problem domains are Consumer Search, Logistics Optimization, Supply & Demand, Content & Personalization, LiveOps, Support Comms, to mention a few.
In our data science team, we e-meet at least once a week and have our knowledge-sharing sessions as well as informal talks. During the knowledge-sharing sessions, we discuss various topics. For example, we can have lessons and experience-sharing chats when putting models into production, cover recent improvements of various of our recommendation engines, or discuss new papers and methods from the machine learning scientific literature. Recently I have also shared my experience in building Bayesian statistical marketing models with pymc. It’s totally up to us to choose what we want to talk about next time!
Sometimes we all gather in Helsinki for workshops, and we hope to organize such events more often when the pandemic is over. ✈️
Also, all our data scientists are extremely talented. I’m quite privileged to work with them! Everyone is friendly. There is no place for big egos. Even when we have different opinions, we talk about them openly, discuss facts and come up with the best solution for the problem at hand.
At Wolt, data scientists work within product teams. How is it for you to work in a cross-functional team like this?
Indeed, data scientists work in cross-functional teams. It’s very helpful for our professional development, as we share our daily activities with different experts: product leads, engineers, analysts, and so on. So, by working together, we learn from each other a lot. For example, I improved my coding skills significantly just by working with our engineers — this is a big plus for my skillset. On top of it, the team culture is really amazing and we are always looking to help each other.
And finally — what do you love about Wolt?
As an employee, I do appreciate the time flexibility and proper work-life balance. Wolt is very supportive when speaking about family. It’s always been easy to dedicate time to my son and deal with unexpected issues that might happen when you have kids. 💙
As a customer, I love how smoothly the app works. Also, customer support is super-efficient. If I come across a problem (not often though), they’re friendly and responsive!