Hilary Mason is a huge name in the data science space, and she has an extensive understanding of what’s happening in this space. Today, she answers these questions for us:

  • What are the backgrounds of your typical data scientists?
  • What are key differences between software engineering and data science that most companies get wrong?
  • How should you measure the effectiveness of your work or your team’s work as a data scientist for the best results?
  • What is a good approach for creating a successful data product?
  • How can we peak behind the curtain of black-box deep learning algorithms?

Below is a partial transcript. For the full interview, listen to the podcast episode by selecting the Play button above or by selecting this linkor you can also listen to the podcast through Apple PodcastsGoogle PlayStitcher, and Overcast.

Curtis: Today we hear from one of the biggest thinkers in the data science space, someone who DJ Patil endorses on LinkedIn for data science skills. She worked at bit.ly, the url shortener, and is a data scientist in residence at venture capital firm Accel Partners, a firm that helped fund some companies you may know, like Facebook, Slack, Etsy, Venmo, Vox Media, Lynda.com, Cloudera, Trifacta—and you get the picture.

Ginette: The partner of this VC firm said that Accel wouldn’t have brought on just any data scientist. This position was specifically created because this particular data scientist might be able to join their team.

Curtis: But beyond her position as data in residence with Accel, she founded a company that’s doing very interesting research, and today, she shares with us some of her experiences and perspective on where AI is headed.

Ginette: I’m Ginette.

Curtis: And I’m Curtis.

Ginette: And you are listening to Data Crunch.

Curtis: A podcast about how data and prediction shape our world.

Ginette: A Vault Analytics production.

Hilary: I’m Hilary Mason, and I’m the founder and CEO of Fast Forward Labs (Please note that Hilary is now the VP of Research at Cloudera). In addition to that, I’m a data science in residence for Accel Partners. And I’ve been working in what we now call data science, or even now call AI, for about twenty years at this point. Started my career in academic machine learning and decided startups were more fun and have been doing that for about 10,   12 years depending on how you count now, and it’s a lot of fun!

Ginette: Something I’d like to note here is there’s been a very recent change: Hilary’s company, Fast Forward Labs, and Cloudera recently joined forces, and Hilary’s new position is Vice President of Research at Cloudera. Now, one thing that Hilary talks to is where the data scientists she works with come from, which is a great example of the different paths people take to get into this field.

Hilary I am a computer scientist, and I have studied computer science. It’s funny because now at Fast Forward, our team only has only two computer scientists on it, and one of them is our general counsel, and one is me, and I’m running the business, so most of the people doing data science here come from very different backgrounds. We have a bunch of physicists, mathematicians, a   neuroscientist, a person who does brilliant machine learning design who was an English major, and so data science is one of those fields where one of the things I really love about it is that people come to it from so many different backgrounds, but mine happens to be computer science.

The people on our team at Fast Forward   typically have a PhD in a quantitative field, such as physics, neuroscience, electrical engineering, and then have, through that, learned sufficient programming skill. One of the jokes I make about my team is that we’re essentially a halfway house for wayward academics in the sense that we can absorb people and teach them to be good software engineers, help them understand the difference between theoretical machine learning and applied machine learning, so we have a long history of practice around that, but data scientists come from a few paths. One of them is this academic path, but I’ve also worked with data scientists who came from a software engineering path who really loved mathematics and have found their way into it that way, and I’ve also worked with folks who came from a social science or philosophy point of view on the world or even humanities who really thought deeply about how data systems function and learned what they needed to know to do the work they wanted to do, and so I don’t think there is one path to data science. We have several.

Curtis: Hilary and her data science company do interesting work on the very edge of the applied machine learning space.

Hilary: Let me give you a little bit of context on what we do at Fast Forward Labs. We really have two sides to our work. One of them is that we do quarterly research reports. So we are doing applied machine learning research. We aim to be six months to two years ahead of what’s well understood in the market, and it is generally algorithmic research that is useful today. So for every report we write, we also have a working prototype that demonstrates that capability or algorithmic advantage on a real data set in a product context. And that allows us to go technically very deep.

Ginette: It’s interesting to hear the inner workings here: after they create a working example of a new concept, they publish their work and show what their new creation is capable of, and sometimes that creation shows the limits of what is possible with a certain approach.

Above is a partial transcript. For the full interview, listen to the podcast episode by selecting the Play button above or by selecting this linkor you can also listen to the podcast through Apple PodcastsGoogle PlayStitcher, and Overcast.


Cover Art

https://pixabay.com/en/safe-money-security-2049948/; Creative Commons 0