Back to blog

2 Years of Data Science at Marshmallow

November 30, 2021
2 mins

Time flies when you're scaling up! I joined Marshmallow in 2019 as the company’s first data scientist. Fast forward two and a bit years and the company has eight times the number of employees and has grown through three different offices (not including my bedroom). In data science, we’ve grown a lot too. As well as growing into a team of five people, the team and I have also matured a lot in how we do data science and, of course, we’re still always learning!

It has been and continues to be a fantastic journey. In this blog, I’ll share with you a few of my experiences and the learnings I’ve picked up along the way.

🔥 A baptism of fire

In my very first week at Marshmallow, I took part in a three-day company-wide hackathon. I worked with front and back end engineers to build and deploy an initial version of a machine learning model to predict potentially fraudulent customers and help the nascent fraud team to prioritise their investigations.

Thanks to everyone’s hard work (and plenty of pizza!) we managed to get a functioning model complete by the end of the week which has been iterated upon ever since.

It was a challenging week but incredibly valuable for getting me up to speed with the company’s data and processes. This experience represents the “move fast” value at Marshmallow — we believe the biggest impact comes from acting on ideas as quickly as possible. I’m looking forward to seeing what we can achieve in the first post-pandemic hackathon!

💪 Taking ownership

Fortunately, over the following months, I wasn’t expected to keep up this pace of deployments. But I was able to deepen my knowledge of the surprisingly interesting field of insurance pricing, allowing me to take on some serious responsibility. This culminated with me leading the rollout of an entirely new pricing model in early 2021.

At Marshmallow, we believe in empowering people to take ownership. So, when I expressed an interest in developing my management skills, I was duly given the chance. I now manage two data scientists, and I love the diverse challenges that this brings.

🌱 Growing pains

Having been at Marshmallow from early on, I was fairly exposed to all the areas in the company. As we’ve scaled, I’ve had to learn to adapt and accept a more structured role within a set team. I’ve also had to give away some of my responsibilities to other new hires so that I could keep progressing in my role (AKA “giving away my legos”). And, I’ve had to adjust to working in a company where I won’t always know everyone (we grow so fast!)

But, scaling up at speed has also enabled me to learn. A lot. Whether it’s discovering the ingenious pipelining framework that our Lead Data Scientist brought to the team (developed by QuantumBlack), or having Marshmallow’s first Data Engineer show me and the team how to build automated deployed ETL processes, I’ve been fortunate enough to learn from great people full of innovative ideas.

🔮 Looking to the future

For me, the most exciting thing about Marshmallow is that we’ve only just scratched the surface when it comes to using data science. There are so many great projects still to come…

We’ve got some huge and rich datasets, like customer conversations, that are ripe for applying interesting machine learning (ML) techniques. We’re setting up an MLOps (Machine Learning Operations) team, opening exciting possibilities from automatic model retraining to live feature stores. We’re moving into new product lines and we’ve recently launched our mobile app, creating loads of fascinating opportunities to drive value with data science. Ultimately we aim to become a market-leading data science team, central to the company’s goal of rebuilding insurance for good!

Looking for your next exciting challenge in data science? You’re in luck — we’re hiring! Take a look at open Data Science roles on our jobs page.