Charles B. Dewitt 
Building a data science team is the most crucial part of any organization and yet there is no right way to do it. There are many designations, roles and responsibilities that fall under this composition and it depends on the organization’s need to decide on the structure of the data science team. Some of the parameters that can be taken into account are the context and stage of growth of each company, the need to set up a data science team, etc.
Building data and analytics capabilities in an organization largely depends on the team structure. It is therefore extremely crucial to configure it correctly to increase operational speed and efficiency.
Building a data science team effectively starts with hiring the right kind of people. Many companies also prefer to train their already existing employees to live up to the expectations of the roles the data science team is expected to perform. Depending on the need for hiring to retrain, the data science team can have three different structures in an organization.
Re-training of the IT department: Hiring a data scientist is not an easy task and leveraging talent already in-house is the best way to meet the need to hire data scientists. Some of the responsibilities such as preparing data, training models, creating user interfaces, and deploying models can be accomplished with the help of people working in the IT department. In-house IT specialists can be trained in skills such as performing predictive analytics, machine learning, and others to perform the not-so-complicated tasks.
Although it may have drawbacks such as limited exposure to machine learning methods and data cleaning procedures, or may require payment for simple processes such as model training, testing or prediction, sourcing data scientists within the organization is one of the best ways.
Integrate computer scientists and data scientists: It includes separate data science and computer science specialists who work on their respective tasks such as preparing datasets, training models, or supporting interfaces and infrastructure supporting the deployed models. The combination of machine learning expertise and computing resources is the most viable option for constant and scalable machine learning operations. While this requires hiring an experienced data scientist, the cost can still be less than hiring a dedicated data science team. It’s the best way to leverage investments in existing IT resources while data scientists focus on innovation.
Recruit specialized data scientists: This comes in handy when it is necessary to build a full machine learning framework rather than managing data. Although it requires the highest cost, all operations, from cleaning data and training models to creating front-end interfaces, are performed by a dedicated data science team. For large organizations, this is the most suitable option, as specialized data science teams can complement different business units of the company.
Once you have decided if you want to retrain, integrate or have a team specialized in data science, it is time to define data scientists according to their roles. There are basically two types of data scientists – Type A and Type B – based on their roles. Type A stands for analytics where a person can make sense of data without necessarily having strong programming skills. They can perform tasks such as data cleaning, forecasting, modeling, visualization, etc. Type B, on the other hand, concerns construction. They have a strong statistical background that can build complex structures such as recommender systems, algorithms and more.
Some of the important roles companies can have under the Type A and Type B data scientist roles are:
CAD and CDO: These are leadership roles and are basically needed to oversee all the other roles that exist in the analytics team. They are commercial translators who bridge the gap between data science and domain expertise, acting as both visionary and technical lead.
Data Analyst: The role involves collecting relevant data and interpreting it. Some of the skills required are R, Python, JavaScript, C/C++, SQL.
Business Analyst: A business analyst handles tasks at the operational level. They are involved in translating business expectations into data analytics and are involved in data visualization, business intelligence, SQL and more.
Data Scientist: They are involved in preparing data, cleaning, using data mining techniques, and solving business tasks. Some of the skills required by them are R, Python, SQL, Hadoop, among others.
Data Architect: Data architects work with large amounts of data and are essential for managing data warehousing, defining database architecture, etc.
Data Engineer: They test and maintain infrastructure components designed by data architects. In most organizations, the roles of data architects and engineers are merged because the tasks they perform are closely related.
Visualization engineers: They may be required to provide data science results in the applications that end users face. In most companies, IT units perform this function, while those with specialized data science roles may have a separate role for them.
