Nakul Verma is the new director of Columbia’s MS in Data Science program. A longtime faculty member and one of the program’s most popular professors, Verma is moving into a broader role, shaping the curriculum, opportunities, and student experience. His perspective is shaped by roles at Amazon, where he developed fraud detection models; the Howard Hughes Medical Institute, where he worked on statistical tools for neuroscience; and eight years at Columbia, where his teaching in machine learning has earned him a reputation for pairing rigor with approachability. 

What got you interested in data science?

I started programming as a kid and loved building things, but what really drew me in was discovering how math could be used to design intelligent systems. At UC San Diego, I took some of the earliest courses in AI and machine learning, and it felt like stepping into a field that was both deeply mathematical and wildly creative.

How does that early curiosity shape the way you think about Columbia’s program and its students?

What excites me is seeing the same drive in our students—a mix of curiosity and creativity. That mindset is what makes them succeed here. It’s also why Columbia emphasizes both theory and application: students need to know how to use today’s tools, but also understand the principles underneath so they can push the field forward.

Why is that balance between theory and application so important?

Tools come and go. A durable education means knowing how to use them and how they work under the hood. I often compare it to cars: you can be a skilled driver without knowing how the engine works, or a great engineer without being able to drive. At Columbia, we prepare students to do both.

The field has transformed in recent years, especially with advances in AI. How have industry expectations of data scientists shifted?

Employers don’t just want technical wizards anymore. They want people who understand context, can communicate clearly, and can work across disciplines. 

At the same time, automation is taking over routine coding, which raises the bar for advanced technical skills. That shift opens the door to more ambitious work: designing complex systems, interpreting results, and weighing ethical and societal impacts. It’s much like when calculators became standard—they pushed us to focus on deeper reasoning. AI is doing the same for data science.

How does your teaching philosophy shape your vision for the program?

I believe growth comes from being slightly uncomfortable. If everything is too easy, you don’t stretch. My classes are challenging by design, but I break material down and connect it to real problems so it feels approachable. The goal isn’t just technical rigor—it’s resilience, creativity, and the ability to approach unfamiliar problems with confidence.

What excites you most about this role?

I’m excited about the chance to strengthen connections—to students, to industry, and across Columbia’s schools. Data science is inherently interdisciplinary, and Columbia is uniquely positioned to bring those connections to life. And those connections are what spark true innovation.

What advice do you have for current and prospective students? 

I have two pieces of advice. First, don’t be afraid to be uncomfortable as you learn. If you’re too comfortable, you’re not really growing—and if you’re completely overwhelmed, you can’t learn either. The best growth happens in that middle space where the problem feels challenging but not impossible.

Second, do what you genuinely enjoy. When you like the work, you put in the time without even thinking about it—and people notice. Don’t chase trends just because they’re popular. Follow the problems that truly excite you. That’s where you’ll do your best work.