Find the "Correct" Direction: Interview with PhD student Lisang

More than a year ago, while playing an online game of Mafia, I received an unexpected message from a nearby player, just 0.1 miles away. As we chatted, I found out that he also attended UCLA. We decided to exchange contact information, and to my surprise, he was a fifth-year PhD student in my major. His name was Lisang (Click to view his Linkedin). What began as a random encounter soon became a meaningful connection. Lisang not only introduced me to his own research but also welcomed me into a broader academic community of fellow PhD students, many of whom graduated from top Chinese universities and share a deep passion for both their work and life.

Though Lisang once served as my reading program mentor, guiding me through topics related to large language models, I never had the chance to fully explore his research and learn where he envisioned his path after the PhD, because at that time, he was still trying to figure out himself. Now that he’s recently accepted a research scientist position at Meta, I saw this as the perfect moment to catch up and delve deeper into his journey, professional transition, and insights he can offer. 

In this email interview, I asked Lisang about his research focus, the day-to-day realities of being a PhD student, and the valuable lessons he’s learned throughout his journey. Whether you're considering graduate school or simply curious about the life of a mathematical researcher, I hope his story offers insight and inspiration.

Lisang's personal website page

Can you briefly describe your research focus and what originally drew you to this area of mathematics?

"My research focuses on designing algorithms to solve optimization problems arising from machine learning. To be specific, there are various tasks in neural network model training. However, training neural networks is usually regarded as difficult due to the huge size of the trainable parameters. To accelerate the training process, I design algorithms that are tailored for the training problems. Before I dived into this field. I mainly worked on numerical optimization and physics models. Later, I happened to realize that machine learning training study is an emerging research field. At the same time, optimization plays an essential role in the research branch. So, I grasped this opportunity and applied my previous skills to this new field."

What position will you be stepping into in your upcoming job, and what will your main responsibilities be?

"I will join Meta as a research scientist. At Meta, I will be responsible for developing business foundation models. Roughly speaking, we will apply large language models into business areas, such as advertisement, recommendation."

What does a typical week look like for you in terms of teaching, research, and other responsibilities?

"Usually, I wake up around 8:30 am. After breakfast, I will write down a TODO list for the following day. I note down the ideas accumulated from the previous research and order them from the most important to the least. I will try out the ideas on the TODO list. During the research hours, I occasionally form new ideas. Those new ideas might lead to my repurposing the research goals. Apart from the research work, I also responsible for teaching. Every Tuesday and Thursday, I teach two programming discussion sessions in the morning and hold office hours in the afternoon. Besides, if time allows, I usually support directed reading programs and other research opportunities for undergraduates. During this period, I will spend 2-3 hours every week reading the materials and preparing the reading goals for the mentee. And we will have two regular meeting hours every week."

Where do you see your research evolving in the next five to ten years?

"My research on machine learning optimization is both theoretical and practical. To be specific, I design algorithms that prove to be efficient theoretically. Besides, these algorithms are also numerically verified via various experiments. My research opens several innovative directions for machine learning algorithms. These theoretical innovations also provide guidance for empirical research."

Was there ever a moment when you questioned whether a PhD in math was the right path for you? If so, what helped you push through?

"Frankly, yes. From a utilitarian perspective, pursuing a mathematics doctorate degree is not optimal for my future career goals - working as a research scientist in a tech firm. This research scientist role requires more computer science background and empirical AI research experience. However, I still feel studying mathematics is fulfilling. Life does not consist of pennies and dimes only. And all I know is that I will definitely regret it in the future if I haven’t spent this creative period of my life in applied math study. Although I've faced difficulties along the way, I still feel that everything has paid off for a sincere lover of math."

In your opinion, what qualities or mindsets make someone well-suited for a long-term career in mathematical research?

"Stay calm in the face of mental hardship. Look further ahead. Be able to clear your mind of distractions while delving into mathematical study."

What are some major obstacles you are facing during your PhD, and what have those moments taught you?

"I often have to work on the project idly, with no one else helping me. Sometimes, the setups are mentally challenging — they can undermine the confidence and sense of accomplishment I’ve built, gradually wearing down my passion. In those moments, passion alone may not be enough to carry me through the valley. To support myself, I create detailed plans to help accelerate my progress. I've learned how to stay efficient even when I'm not in an ideal working state."

What is something you wish someone had told you at the beginning of your PhD journey that would have made a real difference?

"Make sure to take enough time to find the 'correct' direction. Firstly, this direction should be interesting to you. Make sure that you are willing to fight in this direction for years, even without external rewards. Secondly, this direction should be an emerging and promising direction. Usually, a promising direction can accelerate your career development. Once you’ve picked your direction, don’t hesitate to find the right circle that everyone holds similar goals and is supportive of each other."

Hope you enjoyed this interview! Lisang’s journey is a powerful reminder that pursuing what you love, even when it’s not the most conventional path, can lead to both personal fulfillment and professional success. His reflections offer valuable insights not just into the world of mathematical research but also into the resilience and curiosity it takes to thrive in it.