Julia Poska | February 22, 2019
Ananya Sen Gupta’s entire career may have looked very different had she not serendipitously stopped to pet a colleague’s dog one day as a postdoctoral researcher in Massachusetts. The dog’s owner connected Sen Gupta with a marine chemist who was seeking a data scientist like her to make sense of unknown compounds in the 2010 BP oil spill.
“In his signature way of awesome honesty, he said, ’You are perfect for the job because you don’t understand chemistry at all!’” she recalled.
Sen Gupta successfully “fingerprinted” that spill, and has been looking at the environment as a data problem ever since. Today, as anassistant professor of electrical and computer engineering at the University of Iowa, she still lends her computational skills to environmental efforts.
Hear Sen Gupta describe her work in kid-friendly terms.
Sen Gupta helps a colleague in environmental engineering analyze harmful pollutants in the air and studies the spread of disease-causing pathogens with an environmental health professor. With two physicists, she’s developing algorithms to find high energy vents in the Earth’s radiation belts and identifying patterns of particles in the Martian ionosphere.
“I think of myself more as an applied mathematician, honestly,” she said.
While her collaborators see the data through the specific knowledge of their fields, Sen Gupta only learns what she must to develop useful tools. To identify the problem and understand the data, she listens to the experts and takes detailed notes, which she later translates into her own language: mathematics.
She is then able to build algorithms that identify patterns in the datasets, which are far too large for manual processing. Because she does not know what her algorithms should find, they are essentially free from the confirmation bias field experts might carry. Thus, Sen Gupta’s objectivity can add great credibility to a researcher’s findings; recall the marine chemist’s excitement at finding a chemistry novice all those years ago.
“Sometimes not knowing is a good thing, because it leads to discovery,” she said.
Listen to Sen Gupta’s metaphor comparing mathematics to a verbal language.
Environmental pollutants and pathogens tend to have complex boundaries that are difficult to define mathematically. Sen Gupta said applying existing models and equations correctly is a skill in itself, but the nature of environmental research lets her work from scratch, too.
“What inevitably happens is when apply something existing to a new problem, it starts well, and then it hits a ceiling,” she said. “To crack that ceiling I have to invent something.”
She makes the majority of her code for those inventions open source, encouraging further discovery from others who can directly use her algorithms.
Though today she is busy teaching and conducting defense-related research on underwater sonar, Sen Gupta said if she could clone herself, she would devote more time to environmental issues, perhaps those related to climate change.
Since she cannot solve every problem on her own, though, she calls for more interaction between other data scientists and environmental researchers.
Learn how a seemingly aimless conversation about coffee and tea came to inform Sen Gupta’s environmental research.
As she sees it, there is unlimited potential for what problems computer engineering can help solve. But such collaborations cannot occur unless experts in vastly different fields come together.
“I would hope that, not just me, but all the data scientists on campus and all the environmental scientists on campus would basically get together in a local coffeeshop, in some happy hour, just sit down and chat about their pet peeves and hopes and dreams,” Sen Gupta said. “Because that would just lead to so much new science.”