Three roles in Data & Insight at Oda — Part 1: The Data Scientist
In the initial article of this blog series, Nina Walberg introduced the three roles in Data & Insight in Oda: Data Analyst, Data Scientist, and Data Engineer. She gave a general description of how the roles are defined and what sets them apart. To give you a detailed overview of these roles, we interviewed three people working at Oda.
We start with Kate Kuzmina, Data Scientist in Shop Platform. Kate is a neuropsychologist turned into a Data Scientist and since her start has been working on making customers’ order experience a breeze. Let’s dive into this interview to learn more about her work at Oda!
What is your background and what attracted you to Oda?
Data scientists in Oda represent a wide variety of academic and cultural backgrounds. This diversity is valued and perceived as a competitive advantage. We have folks with degrees in economics, computational linguistics, space physics, applied physics and mathematics, nanotechnology, cognitive neuropsychology, artificial intelligence, and other fields. Most importantly, we agree that the possibility to leverage a whole spectrum of different skillsets makes Oda’s data science team powerful, unique, and fun to work in. This goes hand in hand with the shared value of mutual support, cheering, and constant learning. Being an expat female career switcher from what is widely perceived to be a soft academic background, I greatly appreciate our deliberate choice to embrace this complexity and diversity.
I came to data science from the ivory tower of university research, namely psycholinguistics and cognitive neuropsychology. These scientific branches methodologically rely on experiments and consequent statistical modeling of cognitive processes based on collected data. I have always been interested in the objective investigation of how the human condition unfolds and I am convinced that the power of the applied method plays a crucial role in how much one can learn about it. Naturally, data science with its enormous variety of data, powerful methods, high complexity, and great potential to make a real contribution to this world attracted me irreversibly.
All of these exciting facets of data science (and more!) can be found at Oda. The most important factor for me is the people at Oda: genuine, broad-minded, bright, and committed. There is a strong culture of respect, trust, and support in the company. I believe this is the best foundation when building meaningful, sustainable, and cutting-edge products.
How does your day-to-day work look like?
A typical working day has a couple of meetings and focus time. Several times a week, we have team-wide and smaller discipline-wide syncs used for planning and alignment. Data scientists devote half a day per week to professional development. I block my calendar every Tuesday after 11 am because I am doing a course on neural networks.
We have flexible working hours. I greatly value this flexibility and can hardly imagine myself working with a rigid schedule, especially if it is not rationally justified. The organization of working processes, the culture of trust, and the high level of conscientiousness in the company make this flexibility possible. We also have freedom in deciding on which days to work from the office and on which — from home. I enjoy working from home because it gives me an extra hour of sleep. I also enjoy working from the office, because there is always a vibrant and positive atmosphere (and lunch!). People are different and should decide on their own what work setup makes them most productive and content, and at Oda this is a possibility for us.
What tools and methods do you use?
We indeed score a buzzword Bingo here! :) To keep the data science projects on track we use GitHub Issues, to write them — Python with a bit of SQL, to containerize them — Docker, and finally to orchestrate and schedule them — Argo. We also have other ad-hoc responsibilities within our data architecture. From time to time, we build dbt data models and make visualizations in Looker. We also design and run experiments to track the performance of our data products. Thankfully, we receive support from the Data Platform team which continuously strives to empower data science at Oda to make our work as effective as possible. Slack and Google Meet are our main means of communication. For documentation and asynchronous collaboration, we use Quip.
What are you working on right now?
I work in the Recommendations team that is a part of Shop Platform. In short, people in Shop Platform define and build what customers experience at Oda’s website and mobile application. Regardless of the specific scope of each team, the final overarching goal is to create space for life for our customers. In the Recommendations team, we aim to entirely reduce the cognitive effort required to fill a cart. In an ideal world, a personalized box with groceries would magically appear at the doorstep of our customers without them raising a finger. This would save them from repetitive shopping chores and create space for life.
In the last Focus period, we launched personalized dinner recommendations as an experiment for several thousands of our customers. The main goal of this project is to make dinner planning personalized, efficient, and inspiring. We operated using two assumptions:
1) a large portion of bought groceries is intended for dinners;
2) people repeat dinners, therefore they would repeatedly buy the same products together.
These assumptions clearly defined two problems. The first problem is to identify what kind of dinners — if any — a specific user intends to make. From already existing features, we discovered what dinners are the most popular among our customers. These data were used to train two machine learning models enabling detection of dinners in customers’ orders. In addition to helping people conveniently buy familiar products for their dinners, we added a spark of inspiration by presenting products that go well with these dinners. However, this was only the first part of the project problem space.
The second problem is to build a great user experience around these dinner recommendations. This clearly implies close collaboration with UX designers, product managers, and software engineers. We are currently running an experiment with this new feature on a limited group of customers, and the results will define the future of the project. Pre-defined objectives and monitoring goals allow us to estimate the business value of the new feature, however, we also strive to learn as much as possible from this experiment.
How do you collaborate with your team?
Short answer: we collaborate a lot. Long answer: The Data & Insight discipline relies on the embedded organizational model meaning that embedded data scientists like myself are fully immersed in cross-functional teams with software engineers, UX designers, data analysts, and product managers. Everyone independent of their discipline actively contributes to the decision-making processes concerning product-related questions. To some extent, each discipline speaks its own language, however in the Recommendations team, we succeed in creating a space of shared meanings, so everyone understands the end goal and how individual efforts contribute to it. Oda folks are genuinely interested in tech and products they are building, so this helps to connect with each other across disciplines. I enjoy cross-functional collaboration and learn a lot from it.
In the Recommendations team, we are five data scientists. We collaborate and sync often with each other and have a strong feeling of shared ownership of what we build; there is also always room for individual work and deep focus when this is the best way to approach a task. We have regular discipline syncs, weekly one-on-ones with the closest manager, peer programming, peer-reviewing, and professional development sessions.
How was it like to join a scale-up in rapid growth?
Oda keeps a firm grip on the steering wheel during rapid growth, so it adds dynamism and positive vibes to workdays. We have not sacrificed on the thoroughness of the recruitment and onboarding processes, quite the opposite — both have become more defined and matured. The saying “it takes a whole village to raise one child,” can also describe a successful onboarding and how we do it in Oda — everyone should be and is, to some extent, involved in making newcomers feel both technically and psychologically comfortable in a new environment.
The onboarding sessions continue for about three intense and exciting weeks. Current team members conduct onboarding sessions, learn from each other, and rotate. For example, I have attended several technical onboarding sessions to learn more about it from my more experienced colleagues and I hope to be ready to hold the next technical onboarding session. This makes the whole process sustainable, and fun.
In addition to our work responsibilities, we also have regular social events that are cheerful, casual, and low-key. They vary from gathering around a bonfire at a lake in the forest, chilling and swimming in Akerselva, eating pizza at a restaurant, playing board games and watching Tesla AI day in the office, going to floating saunas near the Oslo Opera House, heading together for a short cabin trip, and more. There are also more grandiose social events on the company level, such as an upcoming rave party with DJs from Berlin’s Berghain. Regardless of the format, socials in Oda are inclusive and fun.
How do you think the Data Scientist role in Oda looks like in 5 years?
This is a tricky question. Short answer: I do not know. Long answer: Given the rising complexity of current and upcoming challenges together with the rapid pace of innovation in tech, I assume that a Data Scientist role would transform. I can speculate that increasing needs for automatization of complex processes and interdisciplinary collaboration would define the vector of the maturation of data science as a discipline. Likely, several new more specific roles would emerge from a general Data Scientist role — similarly, for example, to the Data Science Engineer role.
What I am confident about is that data science at Oda would go hand in hand with the latest developments in the field and there will be tons of exciting work as well as high demand for curious, bright, and enthusiastic people in Oda in 5 years.
We hope this interview gave you a better understanding of the Data Scientist role at Oda. Thanks for reading, and check out the other two articles in this series on the Data Engineer and Data Analyst roles.
Thanks to Kristian Sweeney, Nina Walberg, Carl Johan Hambro, Siri Bruskeland, and Fredrik Jørgensen for their help in preparing this article.
