Blog: Building A Great Data Science Team
Data Science as a profession is slowly becoming matured, more defined and much better understood by Senior Management & Executives. Decision makers now have some vague idea or have at least heard of data science. Most companies now have at least 1 or 2 data scientists tinkering away at Proof Of Concepts to show value for further investment into teams and infrastructure.
Most senior leaders/decision makers have not been able to justify the upfront cost of investing in a data science team and infrastructure so they plan to hedge their bets by starting small and hiring 1 or 2 data scientists.The only caution against this approach is that the starting small mindset can backfire if the business does not support the incumbent Data Scientists that they have hired with subject matter experts to help shape the problems they can solve.
On other occasions, Data Scientists are given the wrong infrastructure such as normal corporate laptops to process millions of data instead of servers, lack of access to the data /bad quality data and lack of other experts like dev ops, data engineers or architects.
Data Scientists are sometimes seen as the star players on a football team but you still need a full team for your star players to perform. No one is bigger than the collective team and the efforts of engineers and architects are so crucial in enabling data science activities.
I personally think decision makers need to assess if their business is akin to making gains in revenue, reducing cost or making processes more efficient by the means of data science. I recommend hiring a contractor or getting a consulting firm to help your teams make that assessment and also define a success criteria. If the margins are big enough after the assessment to warrant a new data science function then rock on. If not, alternative approaches would be to use contractors/consultants to implement models or off the shelf tools to build solutions for use cases identified.
Once all these hurdles are out of the way and decision makers have highlighted their projects/business problems with the help of consultants or an individual contributor, it is now time to build a data science team as a business function.
Data Science as a business function is a team of data scientists working closely with engineers, subject matter experts and key stakeholders within the business to implement Machine Learning, Natural Language Process and Deep Learning models subject to business objectives/problems. Like any other business function within an established business, it will have clear goals, success criteria, budget, leadership, governance and project plans.
The rest of this article will describe the personality traits, behaviours, and skills required to build a great team. The skills below do not have to sit within one individual, but rather taking into account the strengths and weaknesses of each member of the team to build a well rounded and high performing data science team as a business function.
A great data science team requires a great communicator/storyteller who is able to communicate to the rest of the business at all levels. They are usually the client facing type who enjoy giving presentations, capturing business requirements and explaining key technical concepts to a non-technical audience. They tend to be extroverts and enjoy people interactions. They are sometimes great at building and fostering healthy relationships with key stakeholders. They tend to be associated as the leader of the pack (data science team) but not always the case. They are closely linked to the “The People Person” type.
This team member is the backbone of the team. Though they might be shy/reserved at times (not always the case), they enjoy number crunching, coding and building great predictive data science models. They enjoy having lots of technical development time to learn new techniques and algorithms. They normally have 3 or 4 side projects going on in their free/spare time. They are always trying new tools, techniques and reading the latest blogs. They love to experiment with new data science models and platforms. They don’t enjoy being interrupted when they are in the zone with headphones on writing blocks of code.
The People Person
A team involves humans who have needs. Each person in a team has a part to play. Sometimes internal conflict arises in a team or with various different stakeholders. This individual is great with people and can foster, build and maintain healthy relationships. They are great at identifying the needs of the team. If you find a “Doer” who is a “A People Person”, this person is a great technical coach. They typically enjoy sharing their knowledge with others and have a lot of patience to ensure the juniors in the team are picking up technical concepts and soft skills needed to be more effective. They are unselfish and tend to have others on their minds.
This is the most critical of all hiring decisions. The senior member of the team, who represents the team, sets the vision and the heartbeat of the team. The leader will be extremely crucial as to whether your data science team as a business function will be successful or not. They will galvanize the troops, set challenging goals for the team and ensure objectives are met. This person needs to be well rounded, approachable and with good knowledge of the data science ecosystem, and if possible the business/domain knowledge. As the famous author John Maxwell says, “everything rises or falls with leadership”. The leader is often the centerpiece and the rest of the team is assembled around the leader to help and support them cover their blind spot/weaknesses.
The Strategic Thinker
This team member is also an integral part of a well-established team. The strategic thinker is always thinking about the betterment of the team and the bigger picture. You can normally find them building road maps for projects, data science capabilities and engaging with the business to set and manage expectations. They flag project issues in advance and help plan mitigation plans. It’s beneficial for them to be commercial minded to always keep the business function’s objectives and goals at the heart of what they do. This person is likely to be a “Talker” as well.
The Social Butterfly
This person is very similar to the “People Person” and normally ends up being the same team member. The social butterfly is the person who loves creating social events for the team. It could be the team dinner, monthly drinks or quiz night. They enjoy socializing with their colleagues. This supports team bonding, helping to ease the tension and stress of having to constantly deliver business objectives.
Every team needs a referee. Most data science functions are normally building models to tackle key business areas or decision engines. This is the only way to justify the initial investment and to ensure substantial returns are gained. The referee ensures that best practices are being followed with documentation, git control and enforces a methodology process flow for all model builds. They keep their other team members in check and ensure there are process flows for adding new users to servers, getting access request to databases and so on.
I have tried to link personality traits with various different activities that need to take place simultaneously to build, grow and maintain a great data science team as a business function. As a decision maker, it is helpful to evaluate the needs of your business and whether your business requires a data science team as a business function, off the shelf data science capability or individual contributors to build models for a specific use case.
If building a team, as hopefully this article has eluded to, not all Data Scientists are built the same. So ensure the job role, expectations and success criteria are clearly outlined during the interview process. Additionally, seek to understand what type of Data Scientist you are hiring and ensure the business culture matches their personality type and the expectation the Data Scientist has coming into your organisation.
I hope this article has given you more insights into building, growing and maintaining a high performing data science team as a business function.