Teamwork in Data Science

Back in July 2019, I had the opportunity to work with a group of Data Scientists on a real-world project with New Light Technologies. The project was “simple,” emphasis on the quotes. It involved data gathering and displaying the results. Pretty straight forward. None the less our team gathered to talk through the project discovery, logistics, workflows, processes, etc.… I quickly found out we needed somewhere to organize ourselves. to the Rescue

I’ve used Asana for various projects throughout the different roles I’ve worked in over the years. It’s excellent and integrates with several different services, including most that I use daily — Gmail, Slack, etc.… So why not use it for this project as well.

Asana Didn’t Come to the Rescue

Nothing against Asana, but my assumption was — “My, project management solution is the best and therefore, we’ll begin there.” At first, by the whole group, it seemed this was the ordered, measured, & mutually acceptable next step. It was, however, we were moving fast. Asana would have definitely kept up, but we found ourselves trying to keep the pace of our project and the need also to update Asana. This was becoming a task on having a task.

Slack to the Rescue

We began using slack to get that quick communication back and forth that’s needed in a fast-paced development project. Even in the same room, we were slacking each other. Not that as a “typical” Data Scientist are seen as loners, but it was faster to slack the team, than have to explain to each other and then getting their input, and then responding. Too long.

The Real Process

During our discovery phase of the project, we ALL landed on what was mutually acceptable by all team members communication tool — Slack. Worked great for the team and one on one communication. I would definitely consider slack for future communication tools for a group/team project. I did connect my slack to my asana account and created tasks for me to work on during the project. This integration, for me, worked best.

But what about the Teamwork in Data Science. Well, that happened organically in our group project. It wasn’t evidently clear at the onset, but within our team, we had a Python Beast, an API Wrangler, and a Project Manager (a Data Scientist in her own right) that kept us on the straight and narrow.

Discover & Learn to Work Together

We all worked the discovery portion and brought different views on our given topic, each with a diverse background and take on the subject matter. Wouldn’t that be an issue? So many opinions? We all placed our questions, concerns, expertise into the “Pool of understanding” — Crucial Conversations. The pool of understanding is the place, within a group dynamic, that allows us to place our expertise, questions, information, and the like, in a safe place. Allowing for the whole to have a better view of the topic/problem at hand than we would all individually. This, however, poses a problem.

For us to work together, we have to overcome the “Me,” in the equation. This was relatively simple in our small group. It’s difficult in larger groups. We had no thought of personal success in the project, but more of a project first mindset. This allowed us to forgo typical power struggles within a group dynamic, be it Team Lead, or “my solution is the best” mentality.

Individual Team Strengths

All three of us leaned on what our strengths are. We ended up working through the project very efficiently. Was it because we spent more time in the discovery phase or perhaps the selection of a great communication tool? It could have been that achieved the overarching end to end solution, but with small, measurable wins along the way that added up to our final project submission? All, really.

In short, what can guarantee team success and ultimately, project success? I have a few suggestions:

  • Research — understand your topic well; don’t rush to begin, but rush to learn more, and when done, learn some more.
  • Communication — find the best way for your team to talk to one another online and offline. A place to store shared files, documents, datasets.
  • Humbleness — not all of us are natural-born leaders (e.g., taskmasters) and if we are, that’ll be a mess. Step back and see what you bring to each new team, it can be expertise, support, or organization. You won’t be typecast into the same role unless you’re a ninja at it, then well you’re the expert. But step into those roles humbly.
  • Small bites, not big mouthfuls — Your taskmaster will set you up for the win with small goals allowing for celebratory small wins, leading to the next target, next win, next goal… etc….


Pretty simple, right? It really was, I understand with larger teams, and even small ones, this organic team flow chart won’t typically pan out. You will need to work at it. During our discover stage of the project, we landed on a communication tool and beginnings of a team structure. Though not exhaustive, I hope you’ll find some takeaways within this brief article? Have nuggets to share — please do feel free to share your suggestions on teamwork within Data Science Teams.



Submit a Comment