First, there was just a single IT team, then came along a data analytics team. But given the pace at which data science is evolving, it is high time that companies start looking at building autonomous data science teams to not just gain insight from data but also be able to take action based on it.
Autonomous data science teams are structured to find insights in a self-managing manner. The notion of team autonomy most common in agile methods of operations is to enable teams to make decisions of their own is central to today’s world. Organizations need an Autonomous Data science team today because the need of using data is cross-functional.
“Autonomous data science teams help create holistic solutions to complex business problems and bring these solutions to market to continuously iterate and innovate. The team has in-depth knowledge about the business, defined goals, and algorithms making them an important contributor to solutions and strategies across the organization. A structure of autonomy gives them the liberty to contribute across segments of the organizations while ensuring a creative thought process,” said Kumar Srivastava, Chief Technology Officer, Hypersonix.Ai.
But how important is this autonomy when building a data science team?
Srivastava further explains that for an organization, this co-dependence is of immense value as the nature of AI-powered products means that the core has to be customized to each customer. Building AI products requires a heavy data transformation and ML customization phase. This phase requires immense creativity and patience. Data science teams have to submerge themselves into the problem and be able to rapidly experiment, innovate and push the envelope. In such a situation, hierarchy and gates only serve to slow down progress and impede creativity.
While this may seem like the norm, it is quite difficult for data science teams to attain such autonomy as the freedom itself is not the problem but it is the ability to take action on them that becomes the problem. Most enterprises either struggle to trust leaving a business solution to a data science team or are not adequately prepared to transfer business context, constraints and problems to a data science team. This typically leads to a design-by-committee environment which leads to subpar progress and solutions.
“In many companies, Data Science teams are still the Insights providers and not the action drivers. It is very important to put these data science teams to action drivers because there are loads of insights that will get generated even when you implement them. We should trust the process and enable such data science teams,” said Aabhinna Khare, EVP & Chief Digital Officer, Bajaj Capital.
This is not a one-way street as it takes an equally qualified team to trust them to take important decisions. While it is more than important to know the technical know-how, what makes for an ideal team member, when it comes to an autonomous data science team.
Khare believes that trust is a major part of what it takes to form a functional autonomous team, as the organization needs to know that, let’s say, the insights are not shared before the necessary steps are taken.
Healthy communication and the ability to constantly ask questions and learn about the business problem adds to the skills required. Data scientists should be able to collaborate with other functional teams including product, engineering, marketing, support, customer success, operations, and sales. Data scientists have to work with stakeholders who do not share their technical depth and expertise, so they need to have a high degree of emotional intelligence, patience, and the ability to listen and educate.
An autonomous team does not exist in a vacuum, it operates within the confines of the organization, answering to their Chief Officer, and interacting with the other members of the firm on various topics. The team needs to be codependent with the other branches of the organization so function at its best.
So, how to build a culture that thrives on the autonomy of its data science team?
Ravi Pathak, Co-founder & CEO, Tatvic Analytics said, “Having trust in your team is fundamental to building a sense of autonomy, however, leaders can take more proactive steps to help employees feel connected to their teams and other leaders in the organization. As a leader, you will always have to be available for guidance, help employees create strategic goals, give them the right tools to shine, and give increased flexibility where possible,”
A robust team must be built that has the right levels of seniority, experience, and exposure. A mix of skill sets, backgrounds, experiences, and diversity in the team ensures that the team has access to a wide variety of ideas and thoughts. Clear objective metrics need to be defined and established and communicated to the team. This ensures that success criteria and governing KPIs are clear and understood enabling the team to self-determine their progress.
“Another way leaders can develop a sense of autonomy tactically is by setting employees up with opportunities to grow, develop, and work on special projects. As they create and work on things outside of their immediate job role, their sense of autonomy increases,” says Pathak on different ways to build the right culture in your organization for an autonomous team.
The most important benefit of having autonomy in a data science team is to drive a data-centric culture within the organization and establish a Learn-Plan-Test-Measure process in the other functions. It helps in creating more business leaders who are adaptable/flexible and can successfully drive organizational change.
