IBM’s Data Science and AI Elite team encourages an inclusive and interdisciplinary approach to delivering value to clients
IBM Data Science and AI Elite team members Mehrnoosh Vahdat and Rachael Dottle were just one month into their IBM careers when they received their first assignment last July.
The project jettisoned them into the heart of Africa, where their banking client was looking to surface new business opportunities across the subcontinent. Their mission was to generate a proof of concept designed to enhance the value of data science and inject the results into workflows for business users and clients.
Thanks to the IBM leadership behind the project which promotes an agile AI methodology that emphasizes collaboration, both women found an inclusive, fertile environment that encouraged them to pitch in quickly and even lead the project efforts despite being new to IBM.
Not only were their creativity and skillsets warmly received by the client — but they gained satisfaction in finding a home with a data science team that touts both gender parity and an interdisciplinary approach to data science.
Mehrnoosh is a data scientist whose astute technical mind is accustomed to build machine learning pipelines, including cleaning and organizing data – but she had only worked with other scientists and engineers — never with a data visualization expert to tell a story with data.
Enter Rachael, who specializes in creating such assets to put context and life behind the data to better explain models to business users. Their combined strengths contributed to feature store of data science assets that business users could harness to predict lead opportunities in different industries the bank touches.
Mehrnoosh and the rest of the team used IBM Cloud Pak for Data to develop all models in Python – and then stored them in the platform to reuse across other projects. She also used Auto AI – stretching it to handle even text data – to simplify the data prep and model development. After the models had been generated, Rachael visualized the output to convey the value of the feature store to non-technical users – using animation and simple language to make the case.
“We saw the power of storytelling with data visualization and how it can empower our client to go to the business stakeholder and say here’s the outcome of our project. That was a unique experience,” said Mehrnoosh.
For an increasing number of women worldwide, the IBM Data Science and AI Elite team offers a unique opportunity for them to excel in a career typically dominated by men.
Rachael, herself, has worked on data science teams before joining IBM, but had left the field, doubtful she’d ever return.
“I had left Data Science because I was tired of having ideas that didn’t fit with standard data science methodology and I felt being on a team of mostly men I would get cut off,” she said.
Indeed, the data science field is notoriously dominated by one gender. According to a study of data scientist salaries conducted by Burtch Works, the executive recruiting firm, only 15 percent of data scientists are female. The percentage of women in data scientist manager roles is even smaller—less than 10 percent.
Without a wide variety of stakeholders involved in interpreting the data and building models, biases can easily creep in — and that can put AI projects at risk for failure, or even worse, endangering peoples’ lives or livelihood. But IBM has sought to change the face of data science — citing a large number of qualified women entering into the field.
Worldwide, IBM’s premier data science group has a female representation that is about 75 percent higher than the industry norm in data science with healthy racial and sexual diversity also a marked feature among its ranks. The team has also managed to attract the best and the brightest. Some 20 percent of whom have PhDs with 60 percent holding master’s degrees.
Susara van den Heever, Executive Data Scientist and Program Director, IBM Data Science & AI Elite together with Seth Dobrin, Chief Data Officer Cloud and Cognitive Software recently outlined their approach for going beyond the norm and achieving top notch results, starting by reevaluating the entire hiring cycle — from writing the job descriptions to making the candidate an offer.
With issues of much needed trust, transparency and explainability in AI now on the agenda of most Fortune 500 companies investing in AI, a diverse data science team of equal contributors is sure to increase confidence.
Rachael observes that discussions among all DSE team members have a principal of equity, inspiring more idea generation from more diverse voices in the mix. “It’s a more horizontal feeling and that lets you come up with things that might not be the right idea — but you don’t feel uncomfortable.”
Demystify the challenges of the AI Ladder. Get the report here.