Company: Calico Labs
Title: Chief Computing Officer
Daphne Koller may still be finding her footing as Calico Labs’ new chief computing officer, but the machine learning veteran is determined to build a team that can analyze vast amounts of health data and change the way we think about aging and disease.
Calico, the Alphabet-backed venture helmed by legendary Genentech chief Art Levinson, brought Koller on in August to develop a machine learning and computational biology team. She joined the company from Coursera, an education tech outfit she founded, which now offers massive open online courses, known as MOOCs.
But Koller didn’t get her start in computational biology or in education: Her early work focused on core algorithms, a more theoretical side of computer science. A programmer since age 12, Koller earned her bachelor’s degree in computer science alongside her high school diploma at age 17, and went on to complete a Ph.D. in computer science at Stanford University.
Gradually, though, her interests shifted to work that had a more tangible impact: “Over time, I began looking more and more for real-world datasets where machine learning techniques would make a difference,” she said.
Koller founded Coursera in 2012 because she wanted to change the education world, but after a few years, she missed the science. But rather than returning to academia, she jumped into biopharma, to take on the myriad of challenges of putting academic work into practice. Ultimately, the “only effective way” to transfer tech from academia to industry is to transfer people, and not every researcher wants to be in industry, she said.
At Calico she gets the best of both worlds, Koller said. “The nice thing about Calico is the balance between the ability to do cutting-edge science that breaks new ground on the one side, but also to see it through to actual impact on the lives of patients, which I think is a unique opportunity,” she said.
And at Calico, the biological sciences are seen as “incredibly synergistic,” so the people who create data sets aren’t siloed off from colleagues who do the analysis. “We design it together, work on it together, do the analysis together,” Koller said. “It is a wonderful way to do the science that leverages different points of view.”
Tapping a variety of viewpoints requires diverse teams of people, which can be tough to come by in biopharma, gender-wise. But Koller is used to that.
“If anything, the gender bias is even worse in computer science departments than it is in biopharma,” she said. “It’s definitely something one learns to live with.”
Even when teams are diverse, it takes some extra doing to make sure those points of view are heard, and women have a role to play in that, she said: “It requires a certain willingness to learn to interact in ways that may not come naturally, for example, to assert oneself in a group setting, when it’s 12 men and one woman--you--around a table”.
For young women starting a career, Koller’s advice is threefold: establish the right expectations with your partner regarding duties at work and at home, do not be afraid to pay someone to do certain things so you can spend more time doing what you care about and of course, “Be willing to lean in, to take a seat at the table and let your voice be heard.”
Having only been at Calico for four weeks when she spoke to Fierce, Koller was still getting back up to speed on computational biology. Her biggest priority is prepping her team to receive large volumes of data produced at Calico in-house and with its collaborators.
Koller intends not only to apply existing machine learning techniques to biological data, but also to “think afresh” about developing new kinds of machine learning that can fundamentally change human health. The road is an open one: Because she is taking a data-driven approach, the specific conditions and diseases Calico will tackle will depend on which data are produced and what her team learns from analyzing it.
“I need to build an organization that has the computational chops to really analyze these data and extract maximum insights from them,” she said. “There is so much wealth in insights in data that are currently not being appropriately extracted.”