6 principles that build our Data Scientists’ code

It is not just a matter of collecting, analysing and interpreting data. Check out the six principles that build our data scientists’ code.

a man with glasses holds his a dog on his lap while looking at a screen

Culture is the kind of thing people struggle to define, but they are precise to identify if it’s present or not. In The Culture Code, Daniel Coyle offers the following definition:

“Culture is a set of living relationships working toward a shared goal. It’s not something you are. It’s something you do.”

Usually, it’s not a problem to define and share a goal. But the actions, interactions and relationships to achieve it effectively are harder to shape. And to shape it, companies define values, which are key beliefs that make a whole team act in tune. 

In the end, the company culture will emerge from its values.

As culture is so important Coyle argues it’s the most distinguishing trait of a highly successful team – we’ve created a Culture Committee for the Nubank’s Data Science Chapter

Though Nubank has its own values, which we proudly inherited, the nature of the Data Science work not only demanded new attitudes, but also needed an effort to bring company’s values closer to our context to increase Chapter’s belonging feeling. 

One person at a time

At first, we did what everybody does: we went on an off-site, had a long discussion, and arrived at a few words or phrases that were inspirational, but, after that, we had realized that we’d forgotten everything.

So we returned to Daniel Coyle’s book, and we realized that we had to listen more carefully. Then, we defined a couple of questions to profile our chapter members’ experiences on a daily basis. 

We then interviewed 20 chapter members from all levels and tenures, one at time, creating a very personal interaction to engage people.

The questions were these ones: 

  1. What was the last thing someone did for the chapter you appreciated?
  2. Please tell us about a meaningful interaction you had with a chapter member 
  3. If you would list three characteristics from the data scientists or machine learning engineers you most admire, which  they would be?
  4. Can you describe  a moment when you felt proud of being part of our chapter?
  5. What do you think our chapter lacks and we should improve?

After clustering the answers – we are finally Data Scientists! – in the best possible orthogonal way, we had arrived to six values.

Making it stick 

The values and purpose we came up before – when we did the off site – were not bad, but it was hard to remember because they were too generic. To solve this problem, we knew we would need catchy phrases. And to make it even better, all of them would have to relate to Data Science

We did more than one by value and we had the whole Culture Committee polishing it until we had just one phrase for every value. 

Another issue was that we didn’t want just to say our values; we wanted them all over the place: as Slack reactions, at presentations’ covers and in our swags. So we teamed up with a designer and brought to life our team’s scratches. 

Finally, we had it all!

Bringing values to life

1. Diverse ensembles don’t overfit

We understand the value of different perspectives and experiences. We have a diversity taskforce to ensure everyone’s safety; we adapt our hiring pipelines to bring people with different skills.

2. We reinforce each other’s learning

We are a collaborative team, sharing our knowledge and helping each other to grow as a professional and as a person. Taking time to explain things to others is a continuous project everybody is in. 

3. Mind the person behind the data point

Inside the chapter, see all the members as individuals, a whole person, not only a technician. We share our personal interests, we meaningfully care for each other.

Regarding our models, caring about how they impact people. We investigate our model’s outputs, we take care of the data we feed it.

4. We share the same objective function

Our chapter does whatever it needs to positively impact Nubank. Outside Nubank, we share objectives with the community; we share knowledge via meet-ups and open source.

5. We trust our confidence interval

Our chapter is a safe space where everyone can be who they are, asking questions and sharing vulnerabilities without judgement. 

6. Pursue the global maxima

We are challenging our own knowledge and solutions, we have a burning desire to learn, apply and extrapolate.  We never stop thinking about how to improve current solutions.

After establishing these six values, people started to use these icons and phrases to label activities, announcements, actions, and projects. It became an instantaneous feedback about chapter alignment for every aspect. It reminds us every day about what we care about and makes us happy as data scientists at Nubank.

Enter your name