Published: Sep 22, 2021
Last Modified: Jan 10, 2024
All graduate students eventually hit a point where facing the academic job market becomes inevitable. In my experience, it was intimidating to finally come to the end of my studies and start to reach out to the broader academic community. In fact, I don’t think that the word intimidating does justice to the sense of terror and vulnerability that accompanies one’s first forays in to the job market. However, with the right tool set, support network, and processes, one can make the job search feasible and routine. In this note, I’m going to outline the tools that I used to successfully get a job during the 2020 job season, at the height of the third wave of the Pandemic, with a newborn child at home.
Overall, it took 114 hours and 288 e-mails over a semester1.
Full Disclosure: The job that I got was at the University of Toronto Scarborough Campus (UTSC). I did my doctoral work at the University of Toronto, and had done a lot of sessional work at UTSC. This made me familiar to the hiring committee. However, I was a finalist at a handful of other places and feel that the suggestions I am about to make are generally applicable.
Acknowledgements: A huge thanks to my coach, Rich Furman, for teaching me the process outlined below. Thanks to my colleague Roei Tell for reading this and pointing out various oversights. (By the way, Roei has some great CS Job Search advice.)
When I first went on the job market as a graduate student, my supervisors encouraged me to apply to “as many jobs as possible”. At the time, I applied to about three jobs. That was my limit at the time. I had no idea how many jobs I ought to realistically apply to, or what my chances of success were.
This round, I made up the following heuristic. Suppose that each job has a $99\%$ chance of failure. That is, if the hiring is random, $99\%$ percent of time, you do not get the job. For lots of faculty positions, there are at least a hundred applicants, so this seems like a reasonable assumption2. Things might even be a little bit worse. If you assume this, what is your chance of landing at least one job given that you apply to $N$ jobs?
Well, if you apply to $N$ jobs and the hiring is independent and random, then the probability of not getting any offer is $(99\%)^N$. Thus, the probability of landing at least one job offer is:
$$ P = 1 - (99 \%)^N $$
Luckily, this quantity increases as $N$ increases! How many jobs do you need to apply to in order to have a reasonable chance of landing a job? To phrase is mathematically, how many applications do you need to get $P > 50\%$?
$$ P > 50\% \Longleftrightarrow N > \frac{\log(50\%)}{\log(99\%)} \approx 68.96 $$
So, according to this model, you need to apply to at least 69 jobs.
Sixty-nine applications sounds like a huge and unreasonable number of applications to send out. When I was a graduate student, I burned out after sending about three applications. How was I able to send out more than twenty times as many applications during the 2020 job season? By having a clearly defined process.
For me, a process is a complex of two things: a detailed plan and a mechanism for checking progress on that plan. For the job season, I had a plan about how to find and handle job postings and a spreadsheet of job applications that I was going to apply to. My thought was that if I just stuck to the process, and diligently entered things in to the spreadsheet day-after-day, sending out the relevant applications as needed, then I would eventually apply to a lot of jobs.
The process applying to a single job had four entirely separate parts:
I’ll start off by talking about the spreadsheet, since it is the main mechanism for ensuring progress on the job hunt. My columns of spreadsheet were as follows:
Most of these column headings are self-explanatory. It is worth mentioning that the application due date is the most important column. I sorted by spreadsheet by due date and used it as a calendar of upcoming applications.
The process for adding jobs to the spreadsheet was as follows. I would look at webpages such as MathJobs, HigherEdJobs, and the Chronicle of Higher Education. If there was a new job for the search (I will say more about selecting jobs to apply to later) then I would process it in to the spreadsheet. To do so, I would add the Due Date, Position Title, Documents Needed, and links to the Job Ad, the college’s Math Program, and their Mission Statement. At which point, I’d stop. That’s it! I’d completely lose interest in the position and do no more. That completes the first step of the process.
The second step of the job application process, usually done at a later time, was to look at the spreadsheet and find some schools to scope out. This consisted of looking at the job ad, the school’s math department, and their mission statement and summarizing my findings in the spreadsheet. This part of the process is the most fun, by far. I’d write a short note about what loved about the school in the Love! column of the spreadsheet, and any notes about the school and its vision in the Notes column. If, for example, the school was a Hispanic-Serving Institution (HSI) or had a focus on aeronautical engineering, then I’d note that down. If the campus was near a large nature reserve, I’d note that I love hiking and the outdoors. Once I had two or three sentences about the school, I’d stop and completely put it out of my mind. Second step of the process is done. Moving on!
The third step is the emotionally difficult part of applying to a job. I’ll talk about overcoming negative self-talk and hard emotions later on. Writing cover letters, and adapting materials to each school is hard. It requires genuine creative input, and felt like a emotionally heavy-laden chore. So, I made it as simple as possible by only referring to the notes in my spreadsheet and having thorough templates for all the standard application documents. Each cover letter had a paragraph that could be modified to talk about the things that I love. Each teaching philosophy could be adapted to address the specific needs of the math department. This is the sort of work that copy writers get paid to do. Later on, I’ll talk about getting help with this part of the process. After about a dozen applications, it became routine. Once a document package was complete and ready to send out, I’d note that in my spreadsheet and leave all the files in place.
The last step of the process is the most tedious and menial: sending out the applications. It seems like every school (that does not use MathJobs or Interfolio) has their own kooky application website. Sometimes, the files need to be at most 1Mb large. Sometimes, they only accept files in Word 97 format. On average, it took about a half hour to double check and send off each application. Once an application was sent off, I’d change the Application Status column to SENT and leave it up to fate.
Why is this process so convoluted? Why are there four steps? Why not just apply to the jobs one-by-one? The process is laid out like this to help minimize the cognitive and emotional load of applying to each job. I designed and followed these steps so that each act would feel small and atomic, and I would feel more like a cold robot than a living breathing human being with sweaty palms and a pounding heart. By making a systematic routine, with clear entry points and exit points, we can get rid of a lot of stress while applying to jobs.
I found that if I tried to do all these things at once, then I would run in to a serious wall of stress. It was overwhelming. By doing four tiny and distinct steps, which could be done in short bursts, I was able to go about applying to jobs methodically.
This figure is just counting the number of hours spent looking at various job sites, working through the spreadsheet process outlined above, preparing documents, and sending off applications. It does not account for things like all day interviews, exhausted and desparate worrying about jobs, or anything outside. Given these restrictions, the true estimate of time-on-task is probably closer to 250~300 hours. Thanks to Roei for asking about this detail. ↩︎
My colleague Roei Tell pointed out that $99\%$ chance of failure estimate is not generic. It is heavily dependent on the academic year, your track record, and all sorts of factors external to you. There are internal candidates for jobs, who have a $0\%$ chance of failure. There are people like John Pardon, Terry Tao, and Erik Demaine who were rising mega-stars when they were on the job market. The figure $99\%$ is an arbitrary assumption. ↩︎
Thanks for reading! If you have any comments or questions about the content, please let me know. Anyone can contact me by email.