A Q&A with co-CEO Mike Elashoff

Cornerstone AI is publishing a series of Q&As with team members to provide more information and context on their role at Cornerstone AI, as well as their professional background. This is the first post of the series.

Your background is in biostatistics and data science - can you tell me a little about why you chose to go into that field in the first place?

I started as a biochemistry major and, luckily, I wasn’t very good at it. My parents were both statisticians, and my grandfather. They suggested that I take a statistics class, which I resisted because it was what my parents did. But it felt like a great fit right from the first class. 

I really didn't have much of an idea of what I wanted to do with my degrees at the time. When I was in school, statistics was one of the least popular majors at Berkeley; I had many classes with fewer than 10 students. When I told people I was going to grad school for that, a common question was, “What is statistics good for?” The notion that every single company in the world would have statisticians and data scientists and be advertising the importance of those would have seemed ludicrous.

What inspired you to co-found Cornerstone AI? 

At Medidata, I had access to tens of thousands of clinical trials, and at Project Ronin access to hundreds of thousands of oncology patients and their real-world data. In every role, I encountered the time consuming, but incredibly necessary struggle to make the data useful—to clean it, to structure it, to standardize it—before doing any actual analysis. I didn’t want to keep solving this problem, again, and again, and again. I wanted to fundamentally solve this first for myself but also for everyone I knew in the industry who shared this problem. I talked to Andrew about it, and over many conversations we fleshed the idea out, not just the product and solution we’re solving but also about whether it could be a successful business—and one you’d actually want to work at.

On a day-to-day basis, what does your work look like? What drives you to continue the work each day?

It’s a mix of technical and business. As I’ve moved through my career, I have really made a concerted effort to stay as technical as possible. I get a lot of personal satisfaction from being involved in the algorithm development, being hands on with data, staying connected to the actual problem we're trying to solve. For example, answering questions like how to make the results of AI models more accessible and interpretable.

On the business side, it might include talking to investors or customers, thinking about our long-term strategy, or focusing on the team aspect. We have a small team of really bright people, and that requires an environment that bright people are happy in.

What does the future hold for Cornerstone AI? 

One of the challenges in data quality in cleaning data and making it ready for analysis is that it's hard to know when you're done with that process. One thing we've done at Cornerstone is to actually try and quantify that—how the Data Quality Score starts and ends, with visible progress along the way. Having Cornerstone become a known industry standard for what “data quality” objectively means is the future we’re aiming for. We want companies that are working with medical data to be able to use our software to very quickly, easily, and confidently get the data they need and feel confident in the system we’ve built.

What are you most proud of in your professional and/or personal life?

We have two daughters who are 19 and 23. Talk about an area where you start off learning and not so good at something. I do take some pride in having done okay at that. 

I definitely take some pride in the fact that, at Cornerstone, there are a lot of people I’ve worked with before who have some confidence that being a part of this would be valuable for them as well. Everyone at Cornerstone could have a bunch of different jobs, and I’m very grateful for the fact that they chose to put some faith in what at the start was just an idea.

Going back a few years, the first company I did, Patient Profiles, was entirely self-funded. It was many hours of nights and weekends making that a reality. For it to finally be successful—not just in being acquired by Medidata, but also in being able to make that software accessible—was immensely satisfying.

What are you passionate about outside of work?

 I’m passionate about finding ways to improve math education for elementary, middle, and high school. In school, math is primarily presented as this boring set of procedures you follow for some unknown reason. From my own career, I love being a statistician. I find it fascinating working with data, the math, the methods. Almost none of that comes across in math textbooks. In the long term, finding ways that can be improved is something I’m interested in.

Anything else you want to share?

There’s an interesting dichotomy in data science. In school, almost none of the training is in data cleaning. Yet, the day-to-day work of data scientists is at least half that, if not 70-80%. Much of how to do data cleaning is post-formal education and on the job. One thing we’ve tried to do is to put our many years of experience with this problem into the software. The software is in some way an extension of ourselves and what we’ve learned in our careers about how to do this. It’s a business, but it’s also very personal.