Given all the advances and recognition in AI that Invoca has achieved, it’s time to pull the curtain back and introduce you to one of our super-talented data scientists, Mike McCourt.
After a decade of doing astrophysics research at Stanford, UC Berkeley, Harvard, and UCSB, Mike made the leap from academia to startup life. Read on to learn more about this transition, the surprising parallels between studying galaxies and building AI models, and what excites him most about our groundbreaking Signal AI solution.
What’s your role at Invoca?
I’m a data scientist. As Invoca moves further into building sophisticated AI for marketers, my role is focused on the following three areas:
- I make sure our AI models perform as well as they possibly can, and that we’re making the best predictions we can based on our data.
- I look for ways to describe why our models make the predictions they do. Since machine learning models can be pretty opaque, finding ways to explain and justify our predictions helps us understand them better. This is not only helpful to our customers, but enables us to develop new and better models.
- Finally, I research new AI models to use in the future to provide new and better insights to our customers.
From astrophysicist to data scientist? How did you decide to make that career transition?
I’m absolutely fascinated by astrophysics, and I’m so grateful that I have had the opportunity to study it professionally. But, after ten years of research, I was itching for a change. At the same time, I had watched AI take off as this exciting new field of study. It’s an area where a lot of progress and new discoveries are made every month. And given that this innovation is largely happening in the corporate world, it made a lot of sense to leave academia and join Invoca. It felt like a place where I could work with a really smart team and make an impact right away.
How does research life compare to startup life?
I was admittedly nervous about making the transition from academia to startup life. I’d never been in a traditional working environment and I had no idea what it was going to be like. But it has been awesome so far. I’m really impressed with the caliber of people here and with how the company is run.
Astrophysics is an unusual career, and it can be a bit isolating. Each year, I would consider the field as a whole and try to predict what would be the next “big thing” in my area. I would then do a literature review, develop a new mathematical model, write software to solve my equations, and then run my code on big supercomputers. I’d even typeset my own manuscripts for publication. It was really only after I had a finished product that I’d share it with the world, present it to my peers, and solicit feedback — as a result, I’d have long stretches where I worked almost entirely on my own. I like to work independently, and I’m glad that I learned to be so self reliant. At the same time, it’s more fun and so much more efficient to work as part of a team. It’s been really refreshing to see how work is done here at Invoca.
On my team, every morning we have a meeting where everyone goes around and describes what they did yesterday, what they’re doing today, and whether they’re stuck on anything. If anyone is blocked or stuck, someone else jumps in to help them. Going from astrophysics to analyzing call data was a big jump for me, but it has gone very smoothly because I’ve had great people here to learn from, and everyone has gone out of their way to help me get up to speed. It very much feels like we’re all working together to get the best possible results out as quickly as we can.
Can you talk about the parallels that exist between AI and your background in studying galaxy clusters?
As an astronomer, I’d been doing machine learning for years without even knowing it. When you work in science, you collect data that tells you something about the world — in my case, I’d work with images and spectra from huge telescopes. You want to extract as much information as you can from that data, and build models to help you understand whatever process you’re studying.
The real test in science comes when you make predictions for the future: only then do you find out whether your theory really works! That’s almost identical to what I do now as a data scientist: I take data in the form of call transcripts and CRM data and build models to analyze the content of calls and predict their outcomes. Fundamentally, it’s the same process whether I’m predicting whether a conversion happened on a call or whether I’m trying to understand how cold, intergalactic gas drives star formation and builds up galaxies like the one we live in today.
What excites you about Invoca’s Signal AI technology?
As a scientist, whenever I build a model or make a prediction, I hope that I’m learning something about the world. But I only find out when someone uses the model — without that feedback, making predictions starts to feel a bit pointless. What excites me about Signal AI is how useful it is to our customers. I can come out with a new AI model and our customers quickly put it to use at a rate of over 100 million call transactions a year. I like building things, and I always enjoy the sense of craftsmanship when I get to the bottom of a problem or make something work really well… but it’s so much more satisfying to build something when you know people will actually depend on it!
When you look at the application of AI in marketing, where do you see things heading in the next 3 to 5 years?
I’m still learning a lot about marketing, so my perspective will probably change. But I think the great thing about machine learning and AI is that they automate tasks that otherwise would be incredibly tedious or even impossible. This automation allows you to do experiments faster than you otherwise would, and it enables you to ask and answer questions that would otherwise be out of reach. As a result, AI can help you to learn and to innovate so much faster than you otherwise would. If you try more experiments, you’re going to learn more things, and so I think AI is really powerful for developing new ideas in the field of marketing.
How are people typically consuming AI today? What are some of your favorite ‘everyday’ AI use cases?
I think the most striking thing about AI is how silently it’s taken over parts of my daily life. AI algorithms rank the pages for me when I do a Google search, recommend books for me on Amazon, and guess where I’m driving to provide current traffic information. In all these cases, an algorithm has replaced a repetitive task I used to do on my own, and it’s taken essentially no effort on my part.
But I think AI is even more exciting when it’s applied to research and technological development. Since AI algorithms are so fast at certain tasks, they allow engineers to run experiments that would have been too tedious in the past. And faster experimentation really accelerates innovation. It amazes me that we have self-driving cars on the road now, even if they’re prototypes — that would have been unimaginable ten years ago! I’m excited to see what new things are invented in the coming years.
What are some of your hobbies outside of work?
I really like building things. I enjoy woodworking and have made a lot of my own furniture. I used to build wooden canoes and kayaks as well. More destructively, I also enjoy taking things apart and understanding how they work — whether it’s a kitchen appliance like a microwave or something more mundane like an old fountain pen, there’s fascinating physics that underlies the technologies we use every day. And when you take something apart, you can really understand how clever and resourceful the engineers were who designed them. I’m impressed that we understand electromagnetic and molecular physics well enough to use microwaves to cook our food — but I think it’s a small miracle that we can make them so economically!
I also love recharging in nature. I live walking distance to the huge network of hiking trails in the Santa Barbara front country, so you can often find me there after work or on the weekend.