I finally know where I’m headed after my undergraduate degree: a direct-entry PhD at the University of Hong Kong’s MMLab, researching embodied intelligence. While the memories are still clear, I want to write down what happened along the way and leave a point of reference for others preparing to apply to graduate school.

From deciding to pursue a PhD, to failing my first RA assessment, to research internships and a verbal offer I was excited about—none of this was mapped out from the start. Looking back, the help of several senior students, repeated efforts to reach out, and gradually accumulated research experience all brought me here.

Application Results and My Final Choice

Let me start with the results. These include both PhD programs and research assistant (RA) opportunities; here is where each stood.

Program or Opportunity Result and Progress
PhD at the University of Hong Kong’s MMLab Received a verbal offer from the supervisor and decided to accept; faculty- and university-level interviews still remained
RA at PKU’s main campus (first attempt) Did not pass the assessment
RA at PKU’s main campus (second attempt) Received an RA offer; politely declined after consideration
PhD at Tsinghua’s Institute for Interdisciplinary Information Sciences Received no reply after reaching out by email
PhD at PKU’s main campus Met with a professor and intended to pursue the application further; did not follow up after receiving an offer I wanted

How I Got Here

The Starting Point: From Competitions to the Idea of a PhD

During my first two years at Shenzhen University, I mostly competed, mainly in ACM contests. I won a silver medal at an Asian regional contest and also competed in a robotics competition, finishing as national runner-up. Beyond that, I kept my course grades above 90 and spent some time practicing public speaking, expressing myself, and related skills. I hadn’t really encountered research yet.

Still, out of an interest in computer science and digging into problems, I decided before my junior year that I wanted to pursue a PhD. At the time, I wasn’t restricting myself to direct entry; doing a master’s first and continuing afterward was also an option.

First Semester of Junior Year: Starting Research and Reaching Out

In the first semester of my junior year, I gradually realized that applying to research programs required more substantial research experience. Based on my understanding at the time of graduate-school applications for Shenzhen University students, research experience was the most worthwhile investment, followed by GPA, then competitions. But grades were also a foundation, and doing research was no reason to neglect them.

In September, I officially joined a research project at my university. I was still preparing for my final ACM season then, so research progressed rather slowly. While working on the project, I also gathered application information from Zhihu, the “Green Group,” and PhD application chat groups, gradually filling in the gaps in my knowledge.

By December, the university project was nearing completion, and I began reaching out for RA positions elsewhere, hoping a research internship could lead to an opportunity for recommendation-based graduate admission. I mainly contacted Tsinghua, Peking University, and Shanghai Jiao Tong University, both emailing professors directly and trying to contact students in their groups. I aimed high from the start, but had no papers yet, and none of my emails received a reply.

My First PKU RA Assessment

The breakthrough came through a senior student. He accepted my friend request, talked with me at length about my experience and ideas, and referred me to an assistant professor at PKU’s main campus. That led to my first RA interview and assessment opportunity.

The assessment was very challenging for me at the time. I had one week to complete several tasks:

  1. Using the research group’s framework, replace most of the implementation of a particular diffusion-model paper, run experiments, and write a technical report.
  2. Write an introduction to a particular area of AI safety.
  3. Discuss open-ended topics such as GPT.

I was unfamiliar with the relevant material then, my research skills were still lacking, and the timing coincided with exams. In the end, I didn’t pass. The experience gave me a much more concrete sense of what I needed to work on.

The Turning Point: Meeting YM and Choosing Embodied Intelligence

Although I hadn’t passed, that senior student saw how seriously I had approached the tasks and was still willing to help. He then recommended an internship at Shanghai AI Lab, working on embodied intelligence with another senior student, whom I’ll call YM below.

YM met with me to learn about my situation and clearly said he was willing to mentor and train me. His sincerity really moved me, so I agreed. These two senior students truly were the people who made a difference in my path to graduate school.

The day after I agreed to work with YM, another assistant professor at PKU’s main campus invited me to an interview and offered me an RA position. However, this internship required being there in person and came with no promise of subsequent admission. After learning more, I found that the group had quite a few RAs and that competition to stay on for a PhD was substantial. After weighing things up, I politely declined and decided to focus on research with YM.

Research Internships and Further Applications

Final exams ended close to Chinese New Year, and I couldn’t start in Shanghai yet, so I first visited YM’s research group at the University of Hong Kong, where I also met my eventual supervisor for the first time. From what I learned then, the group’s direct-entry PhD students had very strong backgrounds: almost all came from C9 universities and ranked first in their majors. So I didn’t make this one of my main targets. Still, I had a very enjoyable conversation with the professor, and we got to know each other a little.

At the end of February, after the visit, I went to Shanghai AI Lab and officially began my research internship in embodied intelligence. I still had no papers at that point, but YM encouraged me to work hard and said he would help recommend me for opportunities at Tsinghua Shenzhen and PKU’s main campus as they came up. He later did give me a great deal of support with these things.

In April, I interviewed with a professor at PKU’s main campus. By then, I had accumulated some research results and prepared thoroughly. The professor was sincere and had his own insights into research, and we had a very enjoyable conversation. However, because of the school’s admissions arrangements, all he could guarantee was a place at the admissions camp; I would still have to take a written examination afterward. Since I later received an offer I wanted, I didn’t follow up.

Also in April, I contacted Tsinghua’s Institute for Interdisciplinary Information Sciences, commonly known as “Chayuan,” which I had really hoped to attend, but received no reply to my email. Looking back, I hadn’t established contact beforehand and still lacked sufficiently competitive research results. My takeaway is that if you have a clear target early on, you can learn about the group in advance and seek RA or collaboration opportunities, so both sides can get to know each other more fully.

Settling on HKU

As I continued refining my research, I had another in-depth conversation with the HKU professor in April and received a verbal offer. Faculty- and university-level interviews still remained, but as far as I was concerned, my destination was largely settled, and my application season came to a close just like that. It was less dramatic than I’d imagined—if anything, a little sudden.

As for why I gained his approval, my own feeling is that, alongside my growing research experience, speaking up actively and communicating smoothly during our meetings helped him understand me better. The place I hadn’t dared to set my sights on during my first visit ultimately became my choice—something I hadn’t anticipated at the outset.

Additional Background at the Time of Writing

For reference, here are my grades and experience. This summarizes my situation at the time of writing, especially my research results; I did not have all of these when I first contacted professors. Papers marked “under review” are also recorded as they stood at the time.

Academic performance: GPA 4.16 / 4.5, percentage grade 91.00 / 100, rank in major 4 / 156.

Competition Experience

  • Silver medal, ACM-ICPC International Collegiate Programming Contest Asian Regional
  • National runner-up, RAICOM Robotics Developer Competition
  • First prize, Lanqiao Cup National Finals
  • National gold-medal team, China Collegiate Computing Contest—Group Programming Ladder Tournament

Research Results

  • CVPR: one paper each as first author, co-first author (listed second), and fourth author
  • ECCV Workshop: Best Paper Award, co-first author (listed second)
  • ICARM: a third-author paper
  • ACL: one co-first-author paper and one second-author paper, both under review
  • RSS: a paper under review

Research Internships and Visits

  • Shanghai AI Lab: research intern
  • National Engineering Laboratory for Big Data (Shenzhen University): research intern
  • University of Hong Kong MMLab: visiting student

Honors and Scholarships

  • CCF Outstanding Undergraduate (2023, 99 recipients nationwide)
  • Liyuan Star of Excellence (2023), Liyuan Star (2022, 2023)
  • Pengcheng Scholarship (2023), Top Innovative Talent Scholarship (2022, 2023)
  • First Prize for Outstanding Student Leaders, First Prize for Stars of Innovation and Entrepreneurship (both in 2022, 2023)
  • Also Star of Learning, Star of Public Service, Star of Arts and Sports, the Top Innovative Talent Full-Tuition Scholarship, and others, totaling 18 awards

Project Experience: front-end development for an e-commerce platform; second prize in South China in the WeChat Mini Program Development Competition.

Student Leadership and Practical Experience: captain of the ACM algorithm training team; project lead for Guangdong Province’s “Hundred, Thousand, Ten Thousand Project.”

Looking Back: The Three Things I Most Want to Share

  1. Actively pursue opportunities.Don’t let feeling that you don’t yet meet the requirements stop you from even trying to reach out. Emails may go unanswered and assessments may end in failure, but taking that first step is what gives others a chance to get to know you. Taking every opportunity seriously may also bring unexpected help.
  2. Practice explaining who you are and what you do.Expressing yourself, public speaking, and communicating on the spot are all worth practicing ahead of time. When talking with professors, present your experience and ideas sincerely and as an equal, and listen carefully to their feedback.
  3. Learn to make trade-offs at crucial stages.Competitions, coursework, research, and applications all compete for your time, and you can’t do everything to the highest standard. Work out your most important goal at the moment, then decide where to put your energy. When choosing research opportunities, also consider the research direction, mentoring approach, and how much you will actually grow.