Mike Luk, Ph.D. '13: Finding a Signal in the Noise

Mike LukOn his first day in his Ph.D. program at Brown, Mike Luk met a lab partner who would become one of the co-founders of his AI company and shortly thereafter met the other cofounder of the company, who was a postdoc at Brown at the time. After graduate school, they went off to industry for several years. Mike says “During grad school, you always say that you’re going to start a company someday.” They said just that while at a barbecue several years after grad school. Mike laughs at the recollection, saying “We pretty much started the company on Legal Zoom and paid a minimal fee to register an L.L.C.” He explains that they chose AI because they did data science and physics analytics and so AI seemed to make sense. They started with consulting since there were no start-up costs. The group subsequently bid on – and landed – several projects and quickly built what Mike refers to as “street credibility” by working for big companies like Staples and L’Oreal. They built a website around the successes, and people began to come to them.

Later, they partnered with Amazon Web Service (AWS) and Nvidia who had massive resources but no one to build within that framework for them, so they outsourced that work to Mike’s team to build out their platforms. With this work, they grew the company from 0 to 60 clients over the last eight or nine years, with Mike now a Managing Director of the team. Mike’s role at Deloitte entails running the delivery team and technical execution on all projects, with his team now numbering approximately 100 members.

Mike recently volunteered to help Deloitte build learning journeys around AI for the firm and is now the AI Academy dean. As a startup, they had to first learn things they had no background in and then train their own people. “There was a lot of trial and error,” he says. Deloitte approached him with the proposition to bring his knowledge of AI to everyone given his successful experience training his team and the paucity of in-house expertise in this area. His role as the dean is to oversee much of the content building, who their partners are in academia and to build an external footprint with clients. Mike’s purview is 10 to 20,000 people in AI, with varying levels of experience, from people who don’t have any base of knowledge to those who are advanced. One of the key areas that Mike enjoys is influencing large groups of people in the AI learning space. He says “We do a lot of theory, how to talk about it with clients, how to run projects and then we get down to the real nitty-gritty, hands-on keyboard. The participants run the whole spectrum, from just being colloquially good at it to being able to implement high solutions.”

When asked about the intersection of physics and AI, and how a degree in physics helped him succeed in the AI arena, Mike says that they focus on hiring those with Ph.D.’s in stem fields for their talent. He says “A lot of the stem fields are trained for analytical thinking, methodical ways of working and being able to break down a problem systematically; that’s exactly how we operate. Most of the time the actual crux of the modeling, the algorithms, aren’t that important. It's how you break down the business problem and the thought process of taking a client’s huge problem and breaking it down. The algorithms I can tell you, that’s the easy part. It’s the conceptual aspect of finding value, finding a signal in the noise.”

Mike says that the physics theory and equations in his classes at Brown were so much harder than the mathematics necessary in his day-to-day now, making his work much easier. He says, “It is relatively easy to theoretically understand the problems I face because of the formal foundation (in physics).”

Mike advises current physics students to remember that it’s not about the problem, it’s how you solve it. He says that AI people are different in how they tackle problems and that you must find your own way to break down problems and solve them. “Doing all the problem sets and working through experiments has a lot of value in itself, but that process of learning how you would do it and how you best operate is what you are really learning by doing problem sets and physics in general. It’s not necessarily the exact physics behind it that’s useful for jobs external to academia, but formal physics training teaches you how to best solve the problem. For example, by reading about a resource first and then theorizing, understanding the physics behind it, and then breaking it down in these three ways and working through it. If you do it in that methodical way a million times throughout a five-year physics Ph.D., you get used to doing it. Then, every problem you see in a job is much easier to tackle because you already know a way to break it down so that you can solve it.

There are many experiences from his time at Brown Physics that Mike draws upon in his daily work, primary among which are the lessons he learned from his advisor, Ulrich Heintz. Mike says, “He was a great advisor for me, and a personal mentor for my work at Brown, especially his leadership style of trusting me to do my work and helping me when necessary. I try to bring a lot of that into my leadership style.” Mike says that the acceptance of inquisitiveness is ubiquitous at Brown. “My professors were very accepting of inquisitiveness, and potentially being wrong. Being wrong and venturing to suggest something should be applauded rather than being worried about being wrong. One of the greatest quotes from one of my professors was ‘There are no stupid questions, only stupid people.’ That’s one of my favorite quotes because it taught me that you’re not showing ignorance by asking questions.”

Mike loved attending the colloquia and seminars at Brown and remembers a theory colloquium where he spied a rather famous attendee who happened to be visiting Brown to collaborate with our faculty. Mike says, “I was shocked to see Nima Arkani-Hamed, who revolutionized three or four fields in physics, there in attendance in the audience – and he was taking notes! That made a huge impression on me. It was inspiring to see this towering figure in physics taking notes in a seminar given by another physicist. It resonated deeply with me that he was learning things – I even remember that he was wearing a rugby shirt! These are the types of experiences we hope to provide in the AI Academy. We are trying to create a community of excellence in learning similar to that which I experienced at Brown Physics.”