"If you survive this, then nothing will stop you"

Professor He, you work in the field of optimisation and machine learning. What fascinates you most about this field?
Niao He: I’m mostly interested in the exciting interplay between optimisation and machine learning. We know that lots of emerging machine learning applications are centred around data-driven optimisation. By applying optimisation theory to machine learning, we can develop principled algorithms that converge faster, require less training data, and are more scalable. Take LLM for example: with advanced optimsation techniques, we may be able to train the model with much less textual data and fewer GPU hours.

You did your Bachelor’s in mathematics, then you switched to computer science. How did this come about?
It was indeed a unique and atypical experience. I did my Bachelor’s degree in maths, and for my doctorate I wanted to focus more on applied maths, so I ended up pursuing a degree in operations research and working on mathematical optimisation. During my doctorate, I was able to connect with researchers in computer science and realised that there are a lot of common ground between optimisation and machine learning. So I started to work on the interface between optimisation and machine learning and I began collaborating with people from computer science. That’s how I eventually ended up in the field of computer science.

Can you trace your interest in maths and computer science back to your childhood? Or was there someone who ignited that passion for you?
My interest in mathematics started very early on when I was a kid. I always found maths to be the one subject that I’m most excited about. I enjoyed solving math problems and always felt a sense of achievement when I worked out a complex equation or geometry puzzle. I think this was largely influenced by my mom, as math was always close to her heart. When I started applying for undergraduate programmes, I chose maths almost immediately, because it felt closer to my heart, too.

Was there something or someone that inspired you to pursue an academic career? Did you ever consider going into industry?
Initially, I was open to both. But gradually I found myself more interested in research. I like being creative and working on projects where I have the freedom to explore and can take the time to ponder. I also like the teaching aspect. When I was a doctoral student at Georgia Tech, I was asked to teach a course to undergraduate students. It was the first course that I ever taught fully on my own. At the start, I was so scared because I felt like I was still a student myself. But by the end of the semester, the positive feedback from my students really boosted my confidence in becoming a teacher. So I think it's definitely worth taking some trial experiences before getting into academia, such as teaching a small course, or mentoring a junior group member while still in grad school. It helps to learn what it feels like being a professor and ground one's confidence in academia.

Did your advisor support you on your academic journey?
Yes, definitely. My advisor Prof. Arkadi Nemirovski's support was invaluable to me. He was the one who encouraged me to explore different fields to find my true interests. He connected me with some of the top researchers in my field, which really opened up my horizon as to what good research means. He was always there to lift me up whenever I had doubts about myself. Knowing that someone as experienced and wise as him believed in me made all the difference.

You completed your doctorate at Georgia Institute of Technology, then you worked as an assistant professor at the University of Illinois at Urbana-Champaign. Why did you decide to come to ETH in 2020?
When you become an assistant professor, you realise that it’s not only about becoming successful in research but also about having an excellent platform to support your career. At ETH, we have the privilege of generous base funding for pursuing high-risk projects, something most of the top universities cannot match. Another reason is that ETH Zurich has extremely excellent students; they are creative, motivated, hard-working, and autonomous. I enjoy working at ETH a lot.

Working in such an environment must be rewarding.
Yes! Working with students is the most fun part of being a professor. It’s so rewarding to see how your students grow, and how they become leaders in their field. Beyond teaching, it’s the advisership and the relationship with students that is most important to me.

What was the relocation like for you?
It was quite an adventure because I had never been to Zurich before. It was a special time during the pandemic and I did my interview online through Zoom. I didn’t even have an opportunity for a campus visit because of the lockdown. I arrived around December 2020, when all restaurants and shops were closed due to the lockdown. I had no idea what to expect and I hadn’t talked to my colleagues in person.

When did you meet your faculty peers for the first time?
I met my colleagues in person half a year after I had joined. It was quite strange: you’ve already met them online, but you never get to see them in the real world. The first few months, I was also exploring a lot of the culture here. I’d stayed in the US for ten years and I was used to how people interact with each other over there. After half a year living in Switzerland, I got familiar with the area and the culture and I already had supportive colleagues around me. Everything just worked out nicely. Looking back, I'm happy I made the decision.

Did you experience a big cultural difference?
The systems are similar in terms of teaching and research. The main difference is perhaps on the administrative side. In the US, there is more of a top-down system. At ETH, there is more of a bottom-up approach. You should be the main driver of thinking about what research you want to pursue, what courses you’re going to teach and how you’re going to lead your group. You have more freedom, but this can also be overwhelming and you have to allocate your time carefully.

Was there a plan B, in case this path didn’t work out?
There are lots of times when you face frustration and you feel like things are not going the way you expected. You start to think about whether you are fit for the job or not and you start to question and doubt yourself. But I have never thought about quitting the job or quitting being a professor because this is what I’m passionate about – even if there are some setbacks along the way. During those times, I would usually talk to my collaborators and friends and then immediately realise that it’s just common. When I started to learn maths, I felt like if I was able to conquer the subject, then I could do anything. It’s the same with an assistant professorship. I feel like if I’m able to thrive as an assistant professor, I can do any other job because this is one of the most challenging tasks in the world and requires a lot of multitasking skills. If you can survive this, then nothing can stop you.

What does it mean for you to be a good teacher?
Being a good teacher is very important to me. It’s not necessarily about distilling knowledge to students or showing them the latest advances, but rather about whether students will get inspired by you. Maybe there’s a new research topic they will pick up, a career path they will get interested in or just a glimpse of an idea that pops into their heads. I learn so much from my students. Often, I feel like I learn more from them than they learn from me. The students here are very engaged during the lectures and ask astute questions. This leads to a lot of new insights. Every time I teach, even if it’s the same course, I feel refreshed.

In 2023, you became co-director of the Max Planck ETH Center for Learning Systems. What does that role entail and what do collaborations look like?
The Center for Learning Systems (CLS) is a doctoral programme run by the Max Planck Institute and ETH Zurich. The centre involves a number of ETH and Max Planck faculty members, who co-supervise students. As a co-director, you are trying to coordinate between all the involved parties, including faculty and students, and to make sure that the interinstitutional collaboration is functioning well. A critical aspect of the role is to promote research synergies between MPI and ETH, to foster collaboration and enable researchers to conduct impactful research in ML/AI domain.

You are also part of the ETH AI Center core faculty, and of the ETH Foundations of Data Science. How important is it for an assistant professor to take on additional roles and advocate for these interdisciplinary collaborations?
My work is more on the interdisciplinary side, so it’s sort of natural to contribute to other communities at ETH. Institutions like the ETH AI Center provide an ideal playground for getting to know people outside of your discipline and to find common interests. It’s good to stay connected to relevant research communities and quite natural to get involved. The institute, the department and the whole research community in Switzerland and Europe also encourage collaboration. That’s why there are all these joint programmes, like the ETH AI Center, the CLS programme or the ELLIS network, a machine learning network within Europe.

As a young professor, there are also moments of doubt, I assume. Who do you discuss your problems with here at the university?
We have regular exchanges among assistant professors, which I really cherish. We form a network of friends, more than colleagues, that support and help each other. Also, senior professors as well as my mentors offer invaluable advice whenever I approach them with questions.

Do you have specific advice for doctoral students or postdocs who are thinking about going along the academic path and becoming a professor?
For students who are still undecided about whether they should go along the academic path or into industry, it’s always good to gain some experience first. You can go for internships in tech companies for research experience, if you want to have an idea of what research looks like in the labs, or you can consider academic roles, for example picking up teaching a few courses or supervising students working on their Master’s thesis. This provides you with valuable experience to help you decide what’s a better fit for you. I would advise students to start thinking about this early on, instead of waiting until the last moment before they’re on the job market.

You’ve been to China, to the US and now you’re here in Switzerland. You’re uprooted a bit every time and you leave friendships behind. What was that like for you?
A lot of my friends are also researchers or faculty members, with the frequent conferences and academic exchanges these days, it's actually not difficult to catch up with old friends either physically or virtually, that's one asset the pandemic left with us. Also, both here and in the US there’s a welcoming community. Even though I don’t speak German at the moment, I feel like I can still make lots of friends. People are usually very supportive, especially for someone coming from a different culture.

How do you find a balance between your career and your private life?
When your work is something that you are passionate about, then you always find a balance. Work is part of my life, so I don’t feel like it’s work. I would still encourage students to take some time off and make the most of the beautiful nature here in Switzerland. I like hiking a lot. And there’s certainly a lot more I’m exploring and trying to learn.

What are your current and future research goals?
One of our ambitious research goals is to advance both theoretical and algorithmic foundations for efficient and trustworthy decision-making in multi-agent systems, where there are multiple players involved. These systems often result in highly complex decision-making process. Take autonomous systems, for example. Imagine the future: Everyone is using their self-driving cars on the road. For the cars to navigate effectively, they need to account for other cars in the system, road conditions and many other factors. It's fascinating to explore how insights from optimisation and reinforcement learning can help us make more informed decisions. At the moment I’m focussing more on theoretical understanding of these general-purpose problems. In the future, I’m eager to see these advancements applied in real-world scenarios.