Michał Wrąbel gives us an inside look at his duties as a DELVE Senior Data Scientist, from everyday tasks to technologies used—and how it all helps drive client ROI and revenue growth.
Michał, what do you do at DELVE?
I work as a Senior Data Scientist developing various machine learning models. These models can help predict, for example, which consumers are likely to start resigning from a client’s services, or which recently acquired consumers are long-term prospects. Just knowing that some activities are, in fact, predictable forever changes the way our clients think about the usefulness of data for managing their customer relationships.
Of course, it doesn’t stop there. At DELVE, we believe that projects end only after the activation phase, so we’re there to observe how this type of investment in analytics results in a better customer experience—and higher profitability. This “end to end” approach is our preferred one, although we make exceptions.
What is particularly important to you in this job?
I work directly with clients, which means that my relationship with them is more of a partnership than the classic client-contractor dynamic. In fact, clients often treat us as an extension of their own digital marketing department or a specialized unit for data science projects. And it makes sense, because the solutions we implement are much closer to in-house projects.
This in-house character is particularly strong in the case of long-term clients: Our previous projects are really treated as part of the client’s own IT system. Again, this is a good thing since we often work on them together. (And of course maintenance is always handed over to the client, along with a complete architecture plan and documentation.)
Our goal, always, is to create long-lasting improvement for any organization we work for. DELVE is not here to sell an IT system that would need our maintenance for years and make the client dependent on us. That’s just not our style. For us, projects have a clearly defined end. And if the cooperation goes well, the relationship continues in the form of new projects or phases.
What technologies do you work with?
At DELVE, we consider ourselves “technology agnostic” as we usually work with the technologies already present in our client’s infrastructure. Most often, we use Google Cloud Platform (but we also have projects on Azure and AWS, and DELVE people certified in both). While working with Cloud we mostly utilize Python and SQL languages, although our data science department also includes some R enthusiasts.
AutoML from Google Cloud (called Vertex AI) is gradually becoming my favorite tool. Although I still get to write an xgboost/lightgbm model from scratch using the Scikit-learn framework, I also can see how AutoML speeds up data science projects—and how much easier it is to maintain the model and monitor model/data drift afterwards.
When it comes to marketing technologies, Google Marketing Platform (Analytics, Campaign Manager, DV360, Ads Data Hub) and the Facebook ecosystem lead the way, especially in Europe. However, for US-based clients, it’s important to work with other providers such as Amazon or The Trade Desk.
What does your daily office routine look like?
Our everyday work at DELVE is facilitated by seamless cooperation with Google. One of the great benefits of our close relationship is having access to the latest beta solutions, which we do not hesitate to use. And Google, seeing our eagerness to use the bleeding edge of the technology, is usually happy to provide us access to even more betas. Ultimately, this benefits everyone: Google gets honest feedback on its solutions, and our client receives a modern solution—sometimes before it appears on the wider market—provided it performs well in A/B tests. This is especially important for us, and we test all implemented solutions to make sure that they actually develop the client’s business.
What makes DELVE so different?
I love that DELVE is a young and ambitious company that wants to keep growing and developing. Despite having over a hundred employees in many different time zones, we still feel like a startup.
We have dedicated departments, of course, but it’s incredibly easy to get involved in something unrelated to your current position—which I consider an advantage. It’s possible for a data scientist to grow as a project manager, or help with the sales/tender process if that’s of interest to them. Or they can take a more expert route, which includes developing internal solutions for the whole company, creating training in a narrow specialization, blogging, or running webinars.
Every day, I work with a talented team that’s driven by high internal motivation. For me, that’s a huge strength: few things sap your energy more than working with people who are counting the minutes until 5 PM. Not happening here. 🙂
Michał Wrąbel
Director of Data Science