My first 100 days as a professor

May 23, 2024

NeurIPS submissions are done, grades are entered, and thus my second semester as professor has ended. In contrast to the academic job market which has many online guides, there are fewer resources for being a professor. Basically none focus on the transition to junior faculty. In an effort to fill that gap, here are my reflections and learnings from my first 100 days as an assistant professor.

Everything is an experiment. Professors have extreme freedom in how they set up their research labs and spend their time. This was doubly true for me since my joint-appointment bridges two departments: Computational Precision Health, which is brand new and therefore has no existing department protocol, and Electrical Engineering and Computer Science, which is very established and has wide variance in professors.

A few examples of questions I faced – and decisions I made – in my first 30 days:

  • How should my lab communicate – email list, Slack, or Discord?
  • How should I furnish my two offices, one at UC Berkeley and one at UCSF? How should I spend time between them? What about at Berkeley AI Research (BAIR), which sits in another (nicer) building?
  • Where will lab meetings be? Will we have food and from where?
  • Should I apply for this grant that’s due in two weeks? What about this one due in two months?

Recognizing the overwhelming number of decisions I would need to make, I decided I’d seek out as many “first-time” things in the first 100 days as possible. Some of them — like attending a faculty meeting (day 30) and teaching a class (day 92) — were essentially mandatory, but others came through deliberate breadth-first search.

Other firsts included:

  • Submitting a grant (day 45) and getting rejected (day 87)
  • Reading many management resources and ultimately using ~10% of it, specifically about giving feedback and setting lab standards (day 60 onward)
  • Hiring undergraduate researchers for the lab by screening 140 undergraduate applications, interviewing 15 students, and giving out out 5 offers (day 62-68)
  • Declining a talk invitation because I didn’t want to fly across the country for a 1-hour talk (day 81)
  • Creating email text expanders for common responses to cold emails from students and potential collaborators (day 20) and then throwing them out because the >100 emails/week email volume was too much (day 90)
  • Putting up an FAQ and advising statement on my website (day 77)
  • Setting up a weekly lab meeting for paper reading and presenting (day 99) – although we later switched to choosing one theme and focusing the next 4 weekly papers on that topic

Ultimately, the combination of relying on intuition and adapting quickly to new data has given me much needed guidelines and structure, and so far it seems to be working.

Lean into the collaboration. One of the biggest factors for me when deciding whether or not to join academia compared to industry research was the access to students as a professor. Although new professors can suffer from a cold start problem because you have to build a lab from scratch, I’ve enjoyed working with 2-5 postdocs and senior graduate students whom I knew before as well as 5-7 younger graduate students and undergraduate students in my lab. It’s my favorite part of the job by far.

The transition to senior advisor from first-author driver was painful for me since I thrive on having lots of control of a project including the problem setup, methodology, coding, data visualization, and writing. After an adjustment period, I’ve come to enjoy that my new role widens the range of perspectives and directions of my work: my first senior author paper (draft created on day 90, accepted to FAccT 2024 much later) studies the perceptions of LLMs by different groups in maternal health and how we can use that info to guide LLM design and implementation using qualitative and quantitative surveys. It was a great joy to help steer the project towards the postdoc first author Maria’s vision.

To double down on collaboration, I’ve found it both productive and delightful to schedule multiple-hour meetings for projects at pivotal points. It helps me scratch that itch of getting deep in the weeds of a project. I love the iterated brainstorming and extended discussions you can get into on hour 2 or 3 of a meeting. While probably not sustainable as I grow my lab and professor responsibilities, I am hoping to try very very hard to maintain something similar.

Conquer the three-headed beast. The biggest obstacle to unfettered deep focus time is the fact that professors actually have three jobs. People often complain that professor time is fractured between teaching classes, advising students, and crafting a research vision which may include grant writing. While these goals might get in the way of my dream multi-hour research jams with students, I’m really liking the sporadic parts too — probably due to my shorter-than-average attention span and above-average organizational skills. For example, going from teaching a guest lecture straight to giving feedback on student presentations for my own class helps me become better at both. It’s been a fun challenge to viciously optimize my time and develop a lab wiki, email filtering scripts, and general frameworks to do my job even better.

While the mix of tasks is manageable, getting grants to fund my students was one of my biggest anxieties of professor life. Universities provide three-year start-up packages, so there is a lot of pressure to financially support my lab in a sustainable way. Plus, grant writing differs from research paper writing since the time-scale is longer, it’s less feasible to show the actual results, and examples of successful grant proposals are less widely available than successful research papers.

I’ve found that grant writing is just another skill that you can hone with time, focus, and a good strategy. Things that have helped me so far include applying to small grants before applying to big grants, actively seeking feedback on drafts, and serving on review committees for grants (once you start getting invited). Although I’ve had some early success, I imagine I’ll continue building my grant-writing muscle. After applying to 8 grants in my first 100 days (perhaps too many?), it’s starting to seem less daunting.

Revisit the PhD with fresh eyes. Now that I’m on the other side of the desk, I have a whole new perspective on events from my PhD. As a student, I attributed intentions and subtext to professors’ actions that I now know were probably not true. I emailed professors who never responded, I gave rambling research updates to blank stares, and I stressed about whether or not I was making “enough” progress. It’s much more likely that most of the aloof or antagonistic comments from professors are due more to their overwhelming stack of responsibilities than anything else. My first vacant face at a student (day 75) came 30% from not understanding the student and 70% from distraction by an upcoming grant deadline. Because of this, I find it helpful to replay my meetings or difficult conversations from both perspectives – student and professor – to retain empathy for both sides.

One thing I didn’t realize is how exciting it can be to work with students who are self-motivated, curious, and communicative. In a job with many demands and places to put my energy, it’s incredibly rewarding to invest more time into students who have shown a depth of thought or identified the challenges they’re facing. Reflecting on my own PhD, I have gained a deeper appreciation for my advisor’s attention and feedback, especially when I was stuck or feeling discouraged. Before my PhD perspective slips away, my advice to students would be to provide lots of research context while giving updates, to make explicit requests, and to attribute less malice to professors who are actively excited about research but potentially unsure on how to best help students.

Find the right people and the right energy. Because of the long grinding nature of academia, it has been a priority for me to invest in robust support systems early on. My PhD at MIT had a large and close EECS cohort, but faculty life can be more isolating, leading to burnout and an unhappy life. I met two wonderful new friends at UC Berkeley faculty orientation (day 47), and together we’ve created a 20-person mailing list (day 98) for the new faculty in our year, which has led to beer garden outings, Orange Theory classes, piano playing gatherings, and Survivor watching parties. Although it can be very time consuming, I have also made it a point to serve on the organizing committees for NeurIPS and CHIL to stay connected with academic friends.

At the same time, it’s important to curate the atmosphere you surround yourself with. I find myself getting less value out of Twitter/X compared to when I was a student. I got enormous value during my PhD by getting up to speed on the hottest topics, entering a new community, and increasing the exposure of my research. Now, perhaps because of the larger research context I can see as a professor or because the quality of the website is dropping, I find scrolling Twitter to induce more anxiety and frustration. I pop back in sometimes, but I have started seeking sources of creativity or constructive research discussions elsewhere, potentially in spaces I have to cultivate myself.

It’s a marathon, and it’s also a triathlon. Essentially everything about being a professor takes longer than I expected. One of the most frustrating parts of academia is the extremely long feedback loop. As a PhD student, it was agonizing to wait two months before getting reviews on a paper submission. Now, it’s unclear if a research direction will pay off until years later, making an actual marathon seem surprisingly quick.

Observing more senior professors, I’ve been surprised by all of the ways that my academic career could twist and turn. After all, when I started my PhD, GANs were all the rage. My own interests will very likely evolve over time. In that case, I’m hoping that some things remain true: finding pride in my work, striving to make a meaningful impact, and training the next generation of researchers. Check back on me after 1000 days of being a professor!

Thanks to draft readers Ginny Fahs, Ben Kuhn, Jessica Dai, and Amanda Coston.