First Two Years in Academia

My dream job is still a dream job after 2 years. There have been many joys and agonies, but finally I feel somewhat settled — setting up the experiment testbed, securing some grants and starting to supervise PhD students. Here is some story and what I have been feeling so far.

Grant

Academics (faculty members) need to obtain grant, which comes from industry or government or alike. Grants can be open to everyone who meets the eligibility criteria, open within the institute, or closed, which is based on existing connection between PI and funder. I kept applying for several open ones (e.g., this) and getting just rejections. There were no feedback, because those were lightweight proposals (e.g., 2–3 pages including the proposal body and other parts like summary, objectives etc.), and most of them were for a small amount (e.g., for several servers).

In parallel to those small submissions, I had been preparing for a larger grant (e.g., £391k for a couple of servers and 3 year postdoc), which is the most general national grant in the UK for very early stage academics. It took a lot of effort (8 pages plus many online forms like summary, beneficiaries and objectives, and letters from the head of department and industry), which is why other lightweight submissions preceded. I got a few of past proposals from my friends and had my mentor, a senior academic allocated by the department, review my proposal. The mentor was extremely helpful throughput the process, not only for proposal writing but for interaction with the relevant admins in the department. Since this submission involved far more people than my previous proposal preparations, it also played a role of induction to the department.

The review process was very long. I submitted the proposal on 19 November 2020, and received the reviews on 15 April 2021, for which I submitted the 1-page rebuttals. Preparing the rebuttals took me a whole week, and the mentor was again helpful. The final outcome was notified on 12 July. Fortunately, this grant was successful, so I was extremely happy. However, even after the result is released in the email, the official letter, which also describes important information, such as the latest project starting date and the date I can start using the money for equipment, came on August 17, more than a month later. As other colleagues got the notification much earlier, maybe I was just unlucky. Nevertheless, it was a good lesson learned that a grant takes very long to be useful even after the outcome is released.

One week after receiving the official letter of that grant, another good news came in; on 24 August 2021, grant from Facebook was also accepted, which was one that I was particularly happy because it can be considered as award. There was another good news, my first single-author paper was accepted at HotNets’21 on 20 August 2021. Therefore, summer 2021 was a rewarding time.

Research testbed

Since I’m a computer networking and OS researcher, I need physical machines that run custom operating system kernel and file system image without virtualization. Further, multiple users, within or beyond my research group, must be able to share those physical machines. I thus built a research testbed in which a user can reserve direct access to physical machines. It was largely inspired by UCL Hen.

The testbed has one management node and many experimental nodes. The management node connects to the department network and a private network where experimental nodes are attached. Every experimental machine boots up using kernel and root file system provided by the management node. In the private network, the management node thus runs NFS, DHCP, TFTP, DNS and NAT servers; I used dnsmasq because it provides all of those functionalities except for NFS. The testbed supports Linux and FreeBSD.

On top of that infrastructure, I wrote a tool that runs in the management node and manages experimental nodes. It maintains the testbed state, such as hardware spec and reservation of all the experimental nodes, associates the boot loader configuration with the user who reserved the experimental node, and abstracts hardware operations (IPMI commands), such as power-related commands and serial console access.

Building the research testbed was significantly delayed because of the COVID-19 pandemic. First, I had never been able to access the server room until September 2020 since I joined in January 2020; before I figure out things to do, the department, including the server room, became inaccessible due to the lockdown. Second, the procurement process took very long. I first needed to interact with the vendor to figure out the optimal configuration within the budget, which was quick, but then I needed to go through multiple internal processes, such as allocating a rack space in the server room and getting multiple quotes. These procurement processes took two months in total. Also, due to chip shortage, the lead time of the new servers was (and is still) very long. For example, one server ordered on 14 July 2020 arrived on 23 September. Going through the internal process was particularly difficult, because I was not used to it and tended to issue frequent intermediate questions, each of them took hours or day to be answered.

University Duties

Academics have a number of duties, such as teaching (preparation) of regular or noncredit courses, thesis supervision and second-marking of undergrad and master’s students, personal tutoring, which assists undergrad students for their study progression and tutoring, which runs several tutorial-style sessions as a part of another lecturer’s course. I realized that, although individual duties are manageable, they together took up a significant amount of effort. The COVID-19 pandemic particularly impacted new academics. Many of the duties are not adequately documented for new academics. I guess usually those are complemented by casual conversation with colleagues in the department, but working from home made it difficult to both know people and have such conversation. This year was significantly easier than the last year, simply because I got used to the duties and people started returning to the office, meaning that I can solve small questions just by knocking the door of my colleagues. Also, it was so rewarding when I saw positive feedback for teaching. I am thus confident that I can do those duties better and more efficiently in the next years.

Advising and Doing Research

So far I have recruited two PhD students, started in January 2021 and the last month, respectively. Further, one masters’s student has just started working with me, and one postdoc will join in April 2022. Similar to mentoring interns when I was in industry, working with students is definitely exciting, because their work where they get their hands dirty always navigates and stimulates me. However, being the principal PhD supervisor in academia, I feel more responsibility and privilege, and that their long term growth and success are of the most importance. I asked a couple of my academic friends about how they supervise their students, and their words touched my heart.

Advising students is undoubtedly exciting, but I also like doing research by myself, meaning that I lead the project by the most coding and experimenting on the team. I realized that playing such a role alongside university duties and supervision is challenging, but I increasingly started managing that and fortunately was able to publish some results in HotNets. It was also rewarding that I used my research testbed for this paper.

Life in Edinburgh

I am happiest in my life. Thanks to the language and foreigner friendliness of the city and people, I feel very good social integration. I also noticed that Edinburgh is a city where academics can afford home in the city center and walking distance from the university. This greatly helped me think Edinburgh as my permanent home.

3rd year CS faculty at the University of Edinburgh — https://micchie.net