Skip to content

Ahna Girshick, Senior Computational Research Scientist at AncestryDNA


Dr. Ahna Girshick is a senior computational research scientist at AncestryDNA. As an expert in machine learning, data science, and perceptual neuroscience, her research has been published in high-impact scientific journals, including Nature Neuroscience and SIGGRAPH. Her research has been cited in over a thousand times and her work has been featured in The New York Times, Science News, Rolling Stone, WIRED, and NPR.

Using her advanced visual training and design skills, Dr. Girshick developed interactive music apps for Philip Glass, Björk, and Passion Pit, among other musicians. This creative work was shown at the Museum of Modern Art in New York and the Contemporary Jewish Museum in San Francisco.

Prior to joining AncestryDNA, she was the head of product at Enlitic, a Silicon Valley company which applies deep learning to help doctors more effectively interpret diagnostic images. She holds a master’s degree in computer science with a minor in cognitive science from the University of Minnesota and a PhD in vision science from the University of California, Berkeley.

Dr. Girshick graciously gave 30 minutes of her time to This interview has been edited for length.

Interview Questions

[] For the scientifically uninitiated, how would you explain your doctoral and postdoctoral work on what you call the “probabilistic brain?”

[Ahna Girshick] Our brains are ingesting information all the time from birth and that’s what makes the human mind amazing. We’re barraged with information. A lot of it is ingested and digested subconsciously—information about your visual world, your auditory world—coming from all the senses.

When you’re on the train in the morning going to work, there are people around you and objects coming toward you. The brain just seamlessly integrates all of that information. It does us a great service by providing this sense of stability that is integrated and whole, even though it comes from many different measurements. Our eyes, ears, and other sensors make those measurements and the brain does the integration.

Take a thermometer, for example. When my daughter is sick, I take her temperature three times and then take the average because we have a somewhat unreliable thermometer. Our brain is doing that, too. It’s reading in lots of measurements that are not perfectly stable, and they can be combined to create something that feels very stable. It would be very jarring if our world were jumping around all the time, so it’s beneficial and efficient to have this sense of stability.

[] So our brain is essentially making an approximation based on past experiences?

[Ahna Girshick] Right. Past experience weighs in heavily, I believe. We use it to our advantage. There’s a whole statistical framework that was the foundation for my research called probabilistic or Bayesian statistics. In the Bayesian framework for perception, we use the variability in sensory measurements combined with our prior expectations about our world to subconsciously create our best guess at any given time.

In the visual world, most of what happens is done subconsciously and we do it extremely well. We aren’t aware of the automatic processing that occurs to see the the moon as white, round, and far away. We just take it for granted.

It’s an amazing, highly evolved capacity that humans have and it’s so computationally cumbersome that there would be no way to do it without taking these shortcuts. These shortcuts help our brain efficiently get the information it needs and creates this stable sense of the world.

[] Wow. That is a beautiful explanation of something incredibly complex and I appreciate you sharing that with me. I want to talk a little bit about your current work with AncestryDNA. How are you applying machine learning to understanding the human genome?

[Ahna Girshick] I’m in the DNA Science group. We’re basically a R&D lab within AncestryDNA, which is part of I am a research scientist with a computational bent. I don’t have a genetics background, but I work with a lot of people with backgrounds in population genetics and bioinformatics. With them, I solve computational problems to improve the human stories we can tell through genetics.

We have a very large customer base now of ten million people; we’re the leaders in direct consumer genomics. You can think about all these individuals as being related to each other in some sense—like there’s this giant family tree of humanity, a really beautiful vision of human interrelatedness for these times.

My job is to use tools from data science, machine learning, and network analysis to help people learn more about themselves, from their DNA and family history.

[] That’s wild. I’m impressed that you’re able to use so many different types of data sources for these analyses. So you’re incredibly accomplished in science, data visualization, and art. You’ve not only helped shape techniques that are at the cutting edge of data-driven medicine, but you’ve also had your artistic work—including the REWORK_app, a collaboration with the team of Philip Glass—displayed at the New York’s MOMA and the Contemporary Jewish Museum. For a woman of your accomplishment, what would be your dream discovery?

[Ahna Girshick] My interests are in using technology to make us more human. How can we use technology to help us connect to each other more? How can we use technology to spark our creativity, to educate ourselves, to create better societies? Technology for the greater good—it’s for humanity at all different levels. It’s for one person and for our whole society. That’s what inspires me.

[] To ensure the benevolent applications as opposed to something that’s done for greed or personal gain?

[Ahna Girshick] Absolutely. And there are insights about what a benevolent application would be that we haven’t come up with. Hopefully, we’ll discover so much more about humanity via technology. We’re adapting to it as we go and learning more and more.

[] I think that’s what’s so interesting about your background: you are making science and data more accessible to people by creating visuals and ways to display massive quantities of information that might otherwise be inscrutable. Does that make sense?

[Ahna Girshick] Yeah, we have to somehow digest it by finding patterns. There’s just a flood of data coming in and increasing exponentially.

[] I want to move on to the demographics of your industry. In all of your work as a data scientist and a machine learning expert, are you typically one of the only women in the room or is the gender distribution more equitable?

[Ahna Girshick] In my education, it was largely male. I was a computer science major and I was the only female in class, or one of two. That was pretty common and I got used to it. I was in various labs for my PhD and postdocs. I think I had four advisors over several years and they were all male and the labs were largely male.

Startups I worked in were also largely male. That’s because even though I’m interested in problems around visual and perceptual psychology, I always had this computational bent. I noticed that when I was more involved in the psychology department, those labs tended to be all female. They had the opposite issue with not having enough males! I’m grossly stereotyping here, but that was my experience.

Now at AncestryDNA, our science group is more than 50 percent female and we have an amazing female chief science officer Biological sciences tend to have more women. Even in our computational group, the gender ratio is pretty balanced and it’s a wonderful experience because we’re very diverse. Not just gender-balanced, but also people coming from all different parts of the world and different academic backgrounds. It feels really welcoming and stimulating. I feel sad that this isn’t more common because it’s great for everyone.

[] In computer science, why do you think that there has been such a historical gender imbalance?

[Ahna Girshick] It’s hard to say. There may be some sort of vicious cycle. You want to be somewhere where there are female role models and when you don’t have women at the top…

When I was young, I looked for female professors and there were only a few. I didn’t always necessarily connect to them. You need a lot of examples around you, so that’s unfortunate and I’d like to think it’s changing, but I don’t know how much it really is.

Also, being a working parent in our culture and keeping up with your work life and your family life is tough. These challenges are very real and really difficult, and simply extending maternity benefits is not enough.

[] Yes, absolutely. We have to carry the babies in our bodies and we have to feed them. Women are also still doing a majority of the housework and the “second shift.” Being a working parent, especially a mother, is a real challenge.

[Ahna Girshick] Yeah, and even if you can figure out the maternity leave and you have the resources for childcare, it doesn’t stop there. You need to raise a responsible human being when you aren’t at the office.

[] Yeah, it’s very hard to strike that balance, I can imagine. With that in mind, what advice do you have for women who are looking to break into computer science, AI, or data visualization?

[Ahna Girshick] Nowadays there are so many resources that weren’t available to me. We didn’t have Coursera and Udacity. They make it so easy to get started in machine learning without having a computer science degree. There are so many online tools for learning about coding.

There are also Kaggle competitions and the Netflix prize. They put out a public challenge and whoever can crack this data problem will earn a prize and the satisfaction of learning how to solve a challenging problem. They have participants of all ages from all over the world. They’re doing it as a learning experience or to put it on their resumes.

None of this stuff was available when I got started, and it’s brilliant. It levels the playing field of opportunity so that you can be in a small village in a remote part of the world, and if you have internet access, you can get involved.

[] Would you recommend getting a degree in computer science or do you think coding will become obsolete as the engines grow in sophistication?

[Ahna Girshick] Generally speaking, the computer science degree is still quite valuable. There’s a level of training at a good university that covers fundamentals that people taking an online course or two might not ever grasp. I’m speaking in broad strokes. There are amazing self-educated people out there, and there are also people that earn a degree for the sake of the degree and don’t learn it very deeply. But if you have a degree from a good program and there are lots of them out there—it doesn’t have to be from Stanford—I think that is still quite valuable.