An Interview with Ph.D. Candidate Xinyun Chen

By Anni Chai

In the month of March, we began our interview series highlighting stories and advice from female researchers and faculty in celebration of Women’s History Month. This week, we’ve had the amazing opportunity to chat with Xinyun Chen, a Ph.D. candidate at Berkeley advised by Professor Dawn Song. Her research lies at the intersection of deep learning, programming languages, and security. Currently, she mainly focuses on neural program synthesis and adversarial machine learning. Xinyun has been a research intern at DeepMind, Google Brain and Facebook AI Research. She has also been a visiting student at National Institute of Informatics.

Read our full interview with Xinyun below!

Q: Could we get a brief background about your journey in EE/CS?

I started learning programming in middle school for fun and for programming contests. Prior to learning coding, I had been learning and playing piano since kindergarten. Programming and playing the piano are different in many ways, but to become better they both take a lot of practice. I then continued with a major in computer science in undergraduate and now I’m pursuing a Ph.D. in computer science. After all, it’s something I really enjoy working on.

Q: What is your research background and what are you currently working on?

I work on deep learning, programming languages, and security. Currently, I mainly focus on neural program synthesis and adversarial machine learning. A practical application of my focus area is autocompleting spreadsheet formulas based on context in real time during the process of writing formulas. I actually worked on it while I interned at Google and its now available in Google Sheets. It can improve programming efficiency for developers, especially for people who may not have professional experience in using these tools.

Q: What are some challenges you encountered in the field, either in general or due to your gender identity? How did you overcome them? We would really appreciate any anecdotes you have to share!

I think there are different challenges at different stages. During the course study or programming contests stage, I felt it was challenging to get started in these technical disciplines. But these technical fields are hard for everyone, not just for women. In research, there are challenges in both coming up with new topics to explore and implementation. For example, when you are not making progress, you have to decide whether there is an issue with the research topic or have you just not come up with the right approach yet. In research, you have to find a way to combine the problem with the tool and it can be hard at times. If you are not excited about the topic at first, then it can be hard to persevere. So you have to ask yourself, “are you really passionate about your field?”. How I overcome these challenges is to not believe bias that other people may put on me.

Q: How do you see the field evolving ever since you first joined? Are there specific changes that you would like to see happen?

In the technical sense, courses in machine learning and artificial intelligence are becoming very popular in school. In the future, I hope to see more machine learning applications in other fields. More specifically, I’m interested in seeing what impact we can make combining machine learning and programming languages.

Culture wise, I am happy to see more efforts in improving diversity, equity and inclusion across many universities including Berkeley. For example, Berkeley has mentoring programs for folks from underrepresented backgrounds. There are still more things to be done, but I hope this issue will be less and less painful.

Q: Did you have any mentors in your journey that made an impact on you?

Professor Dawn Song is definitely a very excellent researcher herself. She is a woman and she is Asian. She also understands the challenges that international students face. I received lots of encouragement and support from her both in the research area and identity-wise.

I was also fortunate to have opportunities to collaborate with people in the industry, from Google to Meta AI to DeepMind. Some people think relationships you build at summer internships are not long-term, but surprisingly all my mentors are very supportive and I still chat with them after the internships have concluded. They are able to offer a lot of help including now with my faculty search. I feel really honored and fortunate to receive that much mentorship from people. Their different backgrounds have also helped expose me to different perspectives. Resources like this are becoming more and more easily available to students at Berkeley, through different research programs in collaboration with industrial research labs. These are opportunities that students can leverage to help them think more about their futures.

Q: Any general advice for young women or non-binary folks interested in pursuing a career in a technical field?

Of course it would be nice if we can avoid unpleasant comments. But if you still hear bad comments, forget about them and put all efforts into yourself. If you are excited about this field, go for it! Everyone will go through challenges in this field, but fortunately now there are also more and more communities and faculty from underrepresented groups to chat with if you encounter problems. There are a lot of people who are willing to give mental support and encouragement if you reach out. Lastly I would say, believe in yourself and go for it.

Q: How did your research experiences differ in industry v.s. academia?

In industry, you have access to some unique resources to help you conduct your research, however they typically expect production impact in a few years. They don’t favor it as much if you are just exploring techniques without immediate performance improvements. It’s easier to do so in academia as long as you demonstrate future potential. Some top research industry groups are more supportive of fundamental research, but to them it’s still better if your research can have some connection to products. In academia there is less pressure to have that.

Follow AWE Berkeley for more!