How to do CS research in a Liberal Arts College Setting?

Over the past few years, I’ve informally mentored a few junior colleagues as they transition from graduate student to liberal artists college professor.  One of the topics that frequently comes up is how to approach doing research with only undergraduate students. So I decided to write this post to highlight some of the ideas that have worked for me:

Create an Undergraduate Research Lab with a Talent Pipeline

First, give your research group a name. Make it a thing. My lab is called JimiLab which is an acronym for “Just Inspired Music Innovation” since we study music technology. This gives the students a sense of belonging: an academic group that they can identify with not unlike a sports team or student organization.

Like a club, students evolve over the course of their time in the lab. That is, when bright-eyed freshmen and sophomores come into the lab, the more seasoned juniors and seniors can help take them under their wings and show them the ropes. This is a three-way win since it helps get young students up-to-speed, older students a chance to develop leadership skills, and a way for you to distribute advising and mentoring responsibilities.

I also make an effort to have the lab be a welcoming place. For example, I regularly take my students to interesting to lectures at Cornell University (2-miles away). We have fun retreats at my lake house during the summer.  I take students on road trips to academic workshops. I connect current students to alums from the lab. It all adds up to creating an environment where students learn a ton and are motivated to contribute to the research mission.

Summertime is the Best Time

90% of the awesome work that we do happen during the summer when the students in my lab can focus full-time on a project. By contrast, during the academic year, talented students get pulled into course work, student organizations, athletics, and social events. These things prevent even the most dedicated students from getting into the flow when it comes to research since research requires uninterrupted blocks of time and clarity of mind in order to make significant progress.

I’ve been very fortunate over the years to have received internal college funds as well as external Research Experience for Undergrads (REU) funding from the National Science Foundations (NSF) which has allowed me to support anywhere between 2 and 10 students per summer.  The NSF REU Supplement is perhaps the best-kept secret in government funding if either you or someone you collaborate with is connected to an existing NSF grant. 

Have an informal Open Door Policy

I don’t like weekly meetings and neither do my students. IMHO, I think that they are often inefficient both to schedule and when meeting.  Instead, I prefer having an open door policy where my students come in and chat with me whenever my office door is open.  Sometimes we rap for five minutes so that we can make a quick decision and let them know about a relevant paper reference. Other times we take half an hour to rip up the 4×8 foot whiteboards that are in my office. And sometimes they just come in to grab a granola bar from the endless supply in my desk drawer. Either way, the goal is to get a student “unblocked” and send them on their way so that can get back to the job do conducting research.

An open door policy can disruptive which is why I carve out time in my weekly schedule to work from home or close my office door when something important or time sensitive and needs my full focus.  

Focus on Workflow

I talk with my students about streamlining our collaborative workflow with powerful tools. These include:

  • Slack for communication. The major benefit here is that I can observe conversations between subsets of my students without getting involved it my involvement is not necessary.
  • Google Drive for documentation.  This includes docs, spreadsheets, survey forms, and space for file storage.  The ability to edit, comment on, and share keeps version control to a simple minimum. 
  • GitHub for code. The is the easiest way to distribute code, do asynchronous code reviews, and make updates to the code base. I make sure that the incoming research students learn “how to GitHub” right away.
  • Overleaf for (latex) paper writing. Real-time collaborative tools like Overleaf and Google Docs are game changers for efficient and effective workflow. 

Once the (free) accounts are created and everyone has been invited to these shared resources, students are pretty good about figuring out what to do and how to move forward.

Becoming More Applied and Less Theoretical 

In grad school, I studied signal processing, machine learning, and information retrieval. These can all be pretty technical and require deep math, linear algebra, and statistics knowledge. It became apparent pretty quickly that my undergrads did not have all of these skills so it would be pretty hard for me to keep doing what I had been doing in grad school.

The good news for me is that I also like to build things and undergrads are much better at doing this. For example, we spent a number of years building a multi-platform personalized radio app called We operated like a startup tech company within a liberal arts college. It was a ton of fun even though it never really paid off in terms of high impact publications. However, I learned a ton about UI/UX, software engineering, cloud computing, project management, app development, and social media. And pretty much all of the students were able to find awesome tech jobs after graduating because they had had such a deep and relevant experience working on an applied project. 

Find Collaborators at R1 Institutions  

If you are located near a big research institution, it can be mutually beneficial for you to develop a deep collaboration with colleagues there. When I started as a visiting assistant professor at Swarthmore (just outside of Philly), I worked with Prof. Youngmoo Kim from Drexel University (in Philly). I barely knew Youngmoo when I arrived but quickly found out that he is an awesome guy and that we shared a bunch of common research interests. He invited me and a couple of my undergraduate students to spend part of our summer with him and his grad students at Drexel. It was great for so many reasons. First, my student learned what it was life was like as a grad student. Second, I got to serve as an informal advisor to a bunch of his grad students and we wrote a few excellent papers together. Third, and most importantly, Youngmoo served as a valuable mentor to me as I made my transition from grad student to professor.

Two years later, the first thing I did when I arrived in Ithaca was to volunteer to give the weekly AI Seminar at Cornell. To do this, I simply emailed a famous professor named Thorsten Joachims out of the blue. After my talk, Thorsten seemed interested in my “cool” research involving music recommender systems and he wanted to know more. I told him that I was working on an NSF CAREER award application but that I couldn’t spend enough money at Ithaca College to meet the minimum budget since I did not have grad students of my own. He suggested that we instead write a collaborative grant proposal. In a way, I was shocked since Thorsten can work with just about anyone he wants to because he is cool and brilliant. 

Our grant was accepted and we were told informally that it was one of the high rated proposal in our division of NSF that year. The reviews highlighted the fact that Thorsten (with his R1 street cred) brought the “intellectual merit” and I (with my liberal arts angle) brought the “broader impact” which are the two main criteria for evaluating NSF grant applications. Since then, we have published a bunch of papers, co-advised a number of undergrad and grad students, received a second NSF grant, and are waiting to hear back about our third (even bigger) grant application. 

The moral of these two stories is that it is important to reach out to people who might be good potential mentors and collaborators. I have learned and benefitted so much from both Youngmoo and Thorsten that I can’t imagine where my academic career would be like without them.