High-performance computing (HPC) has a well-known diversity problem, and groups such as Women in HPC are working to address it. But while the diversity challenge crosses the science and technology spectrum, it is especially acute in areas of HPC where breakthroughs are driven by extracting insights from data. The deluge of data, with the convergence of simulation and artificial intelligence (AI) workloads, and the development of exascale computers, will all increase the opportunities to generate data and derive value from it.
Taking advantage of those opportunities is not just a matter of adding more bodies, although that’s part of the solution. There is also a specific need for a highly diverse workforce to create next-generation data models and design the machine learning (ML) applications that will yield transformative value from the available data. Teams that represent different perspectives are likely to produce more robust ML models that reflect all aspects of a problem. As Trish Damkroger, Intel’s vice president and general manager of extreme computing, says, “Inclusion is the foundation of high performance and innovative teams. We believe that in order to shape the future of emerging fields like data and computational science, we must bring together individuals with a wide range of perspectives, backgrounds, and experiences.”
Research supports Damkroger’s perspective. Teams with diversity across the lines of gender, race, ethnicity, and sexual orientation show higher levels of creativity and produce more innovative solutions, according to Katherine Phillips, a professor at the Columbia University Business School. Phillips has also found that members of diverse teams tend to sharpen their own and each other’s thinking, resulting in more rigorous problem-solving.
Women in Big Data
One group that’s addressing the need to build an inclusive workforce for analytics and AI is Women in Big Data (WiBD). This industry initiative got its start in 2015 at Intel, shortly after the company had established its $300 million Diversity in Technology Initiative. Noting that women were significantly underrepresented in the company’s big data technologies area, an Intel team reached out to some of the company’s big data partners in the Bay Area to see how widespread the problem was.
“Here was this exciting, up-and-coming area, and we had fewer females proportionately than in some of our hardware engineering groups,” recalls Shala Arshi, senior director of technology enabling at Intel Corporation, a former engineer in the company’s Supercomputer Systems Division, and a founder of WiBD. “We wondered whether it was just an Intel problem or an industry problem, and what we could do about it.”
After determining that other technology companies were facing similar scenarios, they convened a planning team with 15 women from SAP, Cloudera, Oracle, IBM, Intel, and others. Today, WiBD has scaled to more than 7,500 members representing 60 companies and universities. It has nearly a dozen chapters across the US, in Europe, and in Latin America. It also has almost two dozen corporate sponsors and partners, ranging from Netflix and Walmart to Hortonworks and the Linux Foundation.
The growth has been driven by grassroots interest, with activities publicized through social channels such as LinkedIn, Meetup, and Twitter “People find us,” adds Arshi. “We haven’t done a big publicity push. The need is there.”
Chapters hold regular meetings and events focused on evangelism, networking, training, and mentorship. For example, a recent technology panel sponsored by the WiBD Northwest chapter packed an auditorium with both women and men who heard from machine learning experts at Nike, Intel, and Tura.io. Speakers discussed hot topics in big data and AI, from data quality to AI in the cloud to innovative use cases, and fielded almost two dozen substantive questions.
Among other speakers, the audience heard Parham Parvizi, cofounder and architect at Tura.io, tout the benefits of a workplace with full gender equality is. Parvizi’s Portland-based startup uses AI to identify patterns in real-time data streams from Internet of Things devices. “We decided early on to have a 50/50 balanced workforce of women and men,” Parvizi told the audience. “It’s been one of the smartest decisions I’ve made. We have a great atmosphere, a well-balanced team, and excellent decision making. Any company I lead in the future will have gender equality.”
They also got career advice from Richa Khandelwal, a software engineering manager whose team delivers machine learning systems and large-scale infrastructure for Nike. Noting the broad range of open source materials and other resources available on machine learning, Khandelwal told career-changers and others who want to move into big data/AI, “Start by finding a problem you can solve with data and AI. Use those resources, and learn what you need to know as you work on solving the problem. Focus on what you can learn rather than what you already know, and don’t sell yourself short.”
Read the source article in HPCwire.