Total 1 Reviews
Kun gathers multiple researchers and interns to work on a same large project. He provides direct hands-on guidance as he designs the system architecture and persists the design principles throughout the project. Different from many senior researchers, Kun writes code by himself. The first paper I have published was Kun's flagship project of the year. He has already written 1K+ lines of code before I joined the project. After publishing my first paper, Kun let me choose the direction to go, systems or FPGA programming. I decided to choose systems according to his advice, and time has shown it was a good choice. Although Kun offered hands-on guidance for my first project, he required me to do the second project more independently. IMHO this is a perfect way of Ph.D. guidance: students graduated from hands-on advisors often lack ability to carry out independent research, and students of hands-off advisors often struggle with the first publication for years. Project weekly meeting + group weekly meeting + occasional 1-1 meeting. On project weekly meeting, everyone reports project progress, and we discuss technical details. The slides and written materials (e.g., email) must be in English but oral discussion is in Chinese. On project meetings, Kun often comment the slides one by one. He not only pay special attention to overall logic (the ordering of slides, the transition logic between slides, etc.), but also picky on technical details. Sometimes he even require us to show the code and comment the code line by line. This helped us to improve code quality. On group weekly meeting, one or two students do a formal presentation in English. The discussion after presentation is often heated. Occasional 1-1 meeting does not have a fixed schedule, typically once a month. Email is returned within one day. Kun is often immediately reachable at his office.
Do not have a clear requirement on the number of papers, but require every paper to be published in a top venue. Kun said, "if you want to publish the paper on a non-top-tier venue, please remove my name from the author list." Do not expect students to work late or on weekends. This is normal for advisors in MSRA. Kun did not expect us to follow a 9am-6pm workday schedule. I often woke up in the noon and went back very late in the evening, and only have two meals per day. Before deadlines everyone including the PI would rush for about two weeks. On the last day before submission deadline of my first paper, Kun and several other researchers worked for a whole night to polish the paper. Regarding graduation time, there is not enough samples of Ph.D. students because Kun is an advisor in joint Ph.D. program. I graduated normally in 5 years. A student who is one year more senior than me also graduated in 5 years. However, there are two other more senior students who graduate 2 more years than expected. I believe this difference is mainly due to research areas. My research area is hotter and easier to publish papers. The two senior students are in a rather mature research field, so publication is not so easy. The average graduation time in USTC CS is 6 years, so, the average graduation time of Kun's students is the same as the department average.
Kun is very supportive of both visiting other institutions and other aspects of students' lives. He promised that every Ph.D. student should exchange to a university abroad for half to one year. He indeed recommended the two most senior students to exchange abroad for one year or more. During the exchange, they published good papers which made them finally meet the graduation requirements. However, the two junior students including me did not exchange abroad because he left MSRA when we were at the beginning of 3rd and 4th years, respectively. Kun is also very supportive of other aspects of student's lives. When I was in the first year of Ph.D. study and taking courses in USTC, I did not have enough time and willingness to continue the research project, and Kun was very nice to let me enjoy the life at USTC. In joint Ph.D. program from MSRA and USTC, students receive salary from both MSRA and USTC. MSRA internship salary is competitive with other Internet companies, and higher than the salary of most labs in universities. MSRA also grants allowances to interns that need to rent a house because their universities are not in Beijing. The university also grants basic salary from the Ministry of Education. Together, these salaries are sufficient to support basic living in Beijing, so most students do not have to search for other financial sources. USTC has a RMB 15,000 grant for each Ph.D. student to attend an International conference. After this grant is used up, students can utilize the grants from the PI in USTC to attend other conferences. Typically, only first author of full papers can get financial support. MSRA does not financially support Ph.D. students to attend conferences.
The research field is very promising as Kun pioneered research on FPGA-based SmartNICs. He established the direction of networking research in our group and acquired resources to build up a cutting-edge experimental platform. This platform gave us a special advantage for high performance networking research, especially the "secret weapon" of Microsoft (FPGA-based SmartNICs). Kun is a well-known expert in this field. During my Ph.D. study, Kun gradually switched research focus from wireless to data center networking, not only because the latter is easier to publish influential papers, but also because it is key to the company's business. Although his research interest has changed over time, for each Ph.D. student the research interest is focused on the same subject. He often told me to focus, and to converge the ideas after the early exploration stage. This is to fight against procrastination. After Kun left MSRA, he had no chance to remind me to focus. If I had focused more on research in the last two years of Ph.D. research (and not spending too much time on hobbies including mining cryptocurrencies and others), and had implemented some of the ideas rather than always exploring better ideas, I may have published more papers. My first research project was proposed by Kun, and he dictated most of the system designs. He even wrote the first 1K lines of code that laid the fundation of the project. Several other researchers, another Ph.D. student, two undergraduates and me are each responsible for implementing a part of the system. I implemented the core part, the compiler, and coordinated with other project members to put all parts of the system together. In paper writing stage, although we contributed technical contents, the paper is mostly rewritten by him because I did not know how to write a paper yet. Kun honestly credit contributions to members when publishing papers.
Students in the same group often have lunch and dinner together. We are good friends. Kun and me are also friends. Kun is a good manager as he is able to gather resources to build platforms and complete large projects that make impact. Although he is rigorous on design principles and technical details, I feel comfortable working with him. Kun often share a lot of deep thoughts on research during 1-1 meeting, group meeting, or dinner. For example, the mindset for systems research, analytical thinking, and the attitude of removing the fake and seeking for the truth. AFAIK, no student drop out from the group before the expected termination date of internship.
Kun is a researcher that I admire, and that I want to become in the future. I'm fortunate to have him as my advisor and to be in a promising research field that he pioneered. I'm also fortunate to work with him in the same company now with more exciting projects to explore.
Yes, I recommend this PI.