Up for the challenge

Science & technology

How one woman built a diverse data science team from scratch

If your colleagues don’t understand your work, then how do you convince them it’s essential to their business? If you ask ExxonMobil data science manager Sarah Karthigan, it requires a lot of persistence and education.

Karthigan was mesmerized by the possibilities that come with analyzing big data. After completing a graduate program in data science at Harvard University, she was empowered by her leadership team to build out ExxonMobil’s first data science team within IT.

Karthigan knew that the key to a successful team was diversity.

“We needed different perspectives, different backgrounds and different interests inside the team if we were going to develop innovative solutions across the company,” says Karthigan.

As a result, the team she assembled is a strong roster of deep talent, with employees specializing in mathematics, biostatistics, physics, finance and more. The varied outlooks they bring to the table allow the team to challenge the status quo – a critical objective when working in a field that is constantly evolving.

Specifically, the team set out to develop new applications for machine learning (a subset of artificial intelligence) that would analyze data and help reveal new insights.

But like all major campaigns pushing for change, the first challenge was to create awareness. Karthigan realized the team needed to educate others of its own capabilities and how they could be of practical use to ExxonMobil’s business lines.

“We were knocking on a lot of doors and telling people about data science, trying to share this new approach to problem solving,” says Karthigan.

Fast-forward two years – and many successful projects and prototypes later – and the team can’t keep up with the incoming requests for their skills. “People are a lot more open to experimentation and want to see what their data is telling them beyond what they already know,” she says.

The interest in machine learning opportunities continues to grow across ExxonMobil. Many business and service lines have begun to embed data science talent into their teams.

“We are fostering a community approach to growing data science across the company,” says Karthigan. “This approach helps harness the collective capabilities of data scientists across
ExxonMobil to realize value as quickly as possible.”

At a high level, the majority of projects that the data science team works on fall into three categories:

Accelerated discovery

Using Natural Language Processing (NLP) techniques, data scientists help subject matter experts such as geoscientists and geophysicists accelerate the discovery of relevant insights from unstructured data (text/images).

Anomaly detection

This approach trains a machine learning model to understand patterns in data, allowing it to pick up on anomalies. This lets teams proactively identify and alert on issues that could impact reliability.

Demand sensing

Data scientists create models to forecast demand using real-time market indicators. This can be used to evaluate purchasing patterns and adapt strategies based on changing demand.

The capabilities of data science and machine learning are limitless. They’re being used in many facets of consumer life – for instance, smart TVs making predictions based on a person’s viewing history and using your phone’s AI assistant to online order your favorite grocery items.

Now, data science is transforming entire industries, including oil and gas, that are presented with or generate seemingly endless amounts of data and need to use it to operate smarter and more efficiently. The adoption of machine learning continues to grow across ExxonMobil, including the addition of data scientists into embedded analytics teams in the Upstream, Downstream and Chemical businesses.

“We can sink or swim in data, and I think data science is that lifeline,” says Karthigan.

Tags:   big datadata sciencemachine learningNatural Language Processing (NLP)
You May Also Like

Explore More