Body

Dave Murgatroyd on ML, Language, and Leadership

Dave Murgatroyd's lessons on leadership

Dave Murgatroyd is a Rice CS alumnus.

Rice University Computer Science alumnus Dave Murgatroyd (B.S. ’99) had completed three internships before a recruiter’s opening question caught him off-guard.

He said, “It was in the fall of 1997 at a Rice career fair, when someone asked what kind of software engineer I wanted to be. I thought, ‘I need to choose? Hmm … maybe accounting software isn’t for me.’”

That question prompted Murgatroyd to look into areas of specialization, exploring a way to blend his love of software with his love of language.

“I’d taken Spanish and Linguistics classes and was also pursuing a B.A. in CAAM. I thought how interesting it would be to combine a very human thing like language and something as powerful and equally fascinating as software. Crafting a specialty integrating the two really resonated with me,” said Murgatroyd.

Once he discovered his niche at the intersection of language and technology, he pursued it with a passion that led him to more courses at Rice and then to the University of Cambridge for a master’s degree in speech and language processing.

“For a while, I would work anywhere I could dive into problems associated with natural language processing (NLP) and the machine learning (ML) that powers it. This was ten or fifteen years before ML got hot.” said Murgatroyd. “Looking back, I can now see my career in three chapters: craft, management, and leadership.

“First, I spent about seven years developing my craft in NLP/ML and software engineering. The technology fascinated me. In my second chapter also of about seven years, I learned how to serve others as a manager, guiding their development and application of that craft. I also learned how to help teams and other managers arrange scope, schedules, and staffing to deliver as much as possible. My third chapter of the last six years has been about giving leadership to broader groups, helping set a realistic vision and establishing an environment where folks are inspired and empowered to pursue it. Giving leadership involves defining where the organization is going and why. Management is about serving individuals and teams in progressing down that path.”

He observed that university students think about how to do things, but not necessarily why. Rice’s CS degree programs adhere to a rapid progression: students learn a different aspect of computers in each course. The rigorous training experience teaches them to learn new things quickly – a valuable skill for software engineers. But it does not allow much time for introspection.

Murgatroyd said, “I find after you’ve had the chance to really focus on something for a while, you start to look outside yourself, beyond your part of the project. I wondered, ‘Whose life is my work making better?’

“At Spotify our mission is to unlock the potential of human creativity. I look for team members who share that mission, who are motivated by their impact toward it in a culture grounded in our shared values.”

He describes ‘problems bigger than yourself’ as those that you cannot resolve on your own. Tackling complex problems as part of a bigger team requires each individual to advance the ideas of others rather than working in isolation, so Murgatroyd encourages engineers to look for organizations that align with their values and are working on the kind of big problems they want to help solve.

“Take the vision of Spotify: a million creative artists living off their art and a billion fans enjoying and being inspired by it. Is that the kind of world you want to work toward? Making that vision a reality can feel far removed from the next meeting on my calendar, but I try to make sure that meeting ultimately serves that purpose.”

In a talk he gave in May 2019, Murgatroyd outlined three characteristics for contributors he calls Mission-Driven Machine Learners. They are engineers who love customers even more than code, love results even more than research, and love values even more than the value of their bank account.

“Those three characteristics are what I look for in teammates who will be pulling in the same direction,” he said. “That is not to say code can’t be beautiful or research isn’t fascinating or that fair compensation isn’t crucial. But especially in machine learning where code, research, and compensation can be easily distorted, I believe it’s ultimately most fulfilling to pursue work that’s part of a larger purpose you believe in.”

As the engineering lead for Spotify’s Personalization Platform, Murgatroyd describes his role as giving leadership to the teams that help power features like Discover Weekly, Daily Mix, Home, and Search. His team’s larger purpose is to enable connections between listeners and artists, improving the lives of both.

Although Spotify listeners can choose their favorite artists, software engineers rarely get to choose everyone they work with. For that reason, Murgatroyd recommends CS students take advantage of opportunities to learn followership in their course projects.

He said, “Being able to work with just about anyone is really key in your career. Followership can be even more important than leadership — having humility about your own ideas and realizing the value of committing to something even if you don’t think that direction is absolutely perfect. Pursuing the second best thing with zeal is still really powerful.

“You can only give good leadership if you prioritize an outcome that everyone can pursue together, rather than your own influence increasing. The end result should be accomplishing the goal of the team regardless of whose idea gets chosen or who is guiding the team.”

His own lessons in leadership and followership began at Rice. As a sophomore, Murgatroyd enrolled in a programming course taught by Ken Kennedy, who had launched the CS department a decade earlier. In his turn as a team leader Murgatroyd built a system diagram to guide his team’s approach.

“I occasionally stumble across that diagram in my papers and am reminded how that course put me on a trajectory to see what team work really looked like. It’s based on alignment which requires trust. Everyone has the same goals and bears the needs of their teammates in mind. The team members keep each other accountable, and when conflicts arise, they embrace the chance to resolve it. Learning how to contribute to the success of the whole team is just as important – if not more so – than your mastery of the technology.”

Follow Dave Murgatroyd on Twitter: @dmurga.