Your Advisor Has Five Impossible Jobs

I learned several unexpected things working as a software engineer at Google between undergrad and starting the PhD. One such thing was how companies split up the responsibilities of “being in charge” into several different job roles. This made it interesting to see how these roles—and more—were all collapsed onto one person in your PhD advisor.

I count five major jobs for a professor:

1. Manager
2. Principal Investigator (PI)
3. Product Manager (PM)
4. Tech Lead
5. Teacher

Let’s walk through a PhD student’s view⁰¹ of these jobs. Each job isn’t impossible on its own, but doing them all well at once might be.⁰²

Manager#

The “Manager” job is someone who is responsible for growing your career. At Google, I had a Manager who was separate from the team’s Tech Lead, and also separate from the Product Manager (PM).

In the corporate world, Managers try to make sure that you’re on the path to getting promoted. This involves making sure you’re doing impactful, visible work with the correct scope, helping you with your performance reviews and promotion packets, and advocating for you to their peers and bosses.

In the PhD, a good advisor should also help plan your career growth. The first decision you’ll make is: what kind of job do you want? A faculty position at an R1 (top-ranked) university is the most challenging job to get, and your PhD there will look different to get there than if you’re looking for a teaching-focused faculty job, or an industry research scientist gig. An advisor that’s good at being a Manager will help you pick your projects, put you on strategic n-th author collaborations, and grow specific skills (e.g., presenting) based on your goals.⁰³

I’ll also lump in all the fuzzy—but important—soft management stuff here, like helping with conflicts, motivation, and difficult situations.

Principal Investigator (PI)#

The term “PI” doesn’t get thrown around much in computer science, except when applying for grants. I hear it more in the harder sciences (bio, etc.)

I think of the PI job as having two main branches:

1. Grant getter. Grants are the lifeblood of an advisor’s lab, because the sweet sweet cash lets them fund PhD students, who do most of the research grunt work.⁰⁴ Getting a grant involves writing enormous specifically-worded documents, likely in collaboration with other PIs. Grants are in hot competition among faculty, and come with different baggage depending on who gives them to you.⁰⁵ Writing and actually getting grants involves deep technical knowledge, some amount of vision, and a big dose of Faculty Metagame⁰⁶ work to strategically impress the funding agencies with plans that are plausible and comprehensible and related enough to the research you actually want to do that it can all be spun up into the same narrative down the line.

2. Marketer. In the way that CEOs do marketing, advisors are constantly giving talks of their students’ work. They give these talks at other universities, companies, funding agencies, and academic meetups (keynotes and panels at conferences). Their talks are a catchy high level vision (their brand), wrapped around meat and potatoes sections of the published work of their lab (research papers). The raw currency of academia is recognition, which makes all these talks vital for getting the word of both their brand and your research into as many brains as possible.

Product Manager (PM)#

In the software industry, product managers work on the high level question of “what should we build?” This is ideally the intersection of business goals (what makes us money?) and customer needs (what do customers want?). During the work, they coordinate between the teams working on the product (like engineering and design and marketing), and help with prioritizing features. I also think of PMs as maintaining a “user-focused” view of the product: when engineers are building something, they’re always taking a step back and saying, “Will users understand this? Will they need this?”

There’s analogs to all this in academia. First, there’s managing a research project. Many people collaborating means different personalities and goals each vying for their own agenda, and in general everyone that’s not the first author trying to avoid doing too much work. In practice this can end up anywhere on the spectrum between authoritarianism and complete anarchy.

More critically, advisors have what I’d call the “PM’s view” of the project. The “users” are other researchers, so they’re constantly trying to take a step back and think: “Will this interest my colleagues in the field? Will it excite them? What else should we change or add to make this project as conference-acceptable and Twitter-viral as possible?”⁰⁷

Also, just as PMs work on the many aspects of a product launch (PR, blog posts, videos, etc.), an advisor agonizes over the high level positioning of the research paper. This means:

• the title
• the abstract
• the Figure 1
• the introduction
• the review rebuttals
• the Tweet
• the talk (slides and words) or the poster

Tech Lead#

A Tech Lead in industry is responsible for helping scope work, designing the system, divvying up tasks to the engineers, being on hand to answer low-level questions, and reviewing code. They’re kind of the “technical boss” and decision maker.

Junior advisors sometimes do this kind of low-level work too. They generally phase out of it by the time their lab has grown and they’re too busy to cook. But early on, advisors have the time not only to help you with low-level fundamental material (e.g., statistics, math, and linguistics), they’ll even write code.⁰⁸

But even outside of implementation details, the “Tech Lead” job has a robust—but quite abstract—industry / academia analog. An advisor will basically never architect your system design for you or review your code. But the “technical” aspects of research involve lots of low-level decisions: stuff like algorithms, optimizers, loss functions, data transformations, task designs, and statistical measures. My experience has been that even senior advisors have a robust ability to help make decisions in these on-the-ground design choices.

Furthermore, there’s a regular research loop at weekly meetings where the grad student reports what they tried and the preliminary results. Then, the advisor has to help determine what to work on for the next week. So rather than decisions about software architecture and code patterns, the research questions span the low-level landscape of the field’s work.⁰⁹ Research projects regularly take big direction swings. In my experience, these are often the advisor’s call, because students are more likely to keep slamming their head into a problem to hope it works.

The PM and Tech Lead advisor jobs may seem similar, because they’re both about getting a research project completed. But they’re different enough they can be done by different people. The “Tech Lead” job is about navigating and completing the research path until you have good results. The “PM” job is about selecting interesting research directions, managing the co-authors, and packaging up the project into a compelling narrative.

Teacher#

Finally, there’s the job role that every non academic thinks professors spend all their time on: teaching undergrads!

Let me just put it this way:

• Teaching well is an extraordinary art.

• Teaching well has basically zero value in the research world. (PhD advisors are in the research world.)

As such, I think professors who teach well do so because of their own passion or interests, or as a result of their institution’s infrastructure.¹⁰ This is one of those things that is a whole job on its own (university lecturer, college professor, school teacher), though with a higher teaching load and more expectations.

Handling Strengths and Weaknesses#

How does this relate to the PhD metagame? You spend so much time interfacing with your advisor, I think it’s helpful to have a mental model of the jobs they’re balancing.

Your advisor is probably great at at least one of the jobs, and also bad at one of them too.

I’m not saying that it’s possible to reduce every complex life responsibility to a Persona-like skill pentagon, but I’m also not not saying that.

Academia is rich in opportunities for collaboration and mentorship (and co-advising). If you find yourself needing more help in one of your advisor’s job areas, a great place to start is asking a more senior grad student or postdoc. I found that in the PhD, grad students love to help out and are rarely asked to by their peers.