It is the start of month seven of my industry PhD position. The past six months amount to a sixth of my contract’s regular duration, which if voiced this way seems like a lot of time already. If I had to briefly summarize the results of the first six months for people outside the domain it would go something like this: I have started working on a concrete improvement to an existing approach which tackles a problem that seems worth solving. This might sound like I slacked off during the first five months and three weeks, but it’s actually a notable outcome because at the start I only had my supervisors’ vague ideas and completely lacked domain knowledge. From there I have:
- … studied the essentials of computer networks and data centers, revising some basic undergrad knowledge and learning more about the topic,
- … had discussions with coworkers, my supervisors and external researchers to define the overall scenario we’d like to consider (Inter-DC WAN network maintenance as part of the IPCEI-CIS research project),
- … done an extensive literature search to identify the sub-problems that researchers have identified and attempted solving, and the challenges the remain open,
- … presented my initial findings to fellow researchers and colleagues to collect some feedback and further specify the deal of my research,
- … selected a concrete problem (fine-grained distributed traffic engineering in large-scale networks) that I’d like to solve by extending an existing codebase from related work, and which I can imagine becoming my overall PhD topic,
- … met a large number of interesting fellow researchers and forged initial connections that might become useful later on,
- … started a bunch of side projects (supervision, paper writing etc.) that cover related topics and might yield useful insights for my main research branch or some paper co-authorships.
Of course, this does not mean that the only thing that remains to be done is coding. While working on the baseline codebase I will probably discover quite some conceptual and practical issues that need consideration before I can solve my problem with a derived algorithm. Furthermore, I will most likely keep refining my overall problem setting with the help of fresh insights, just like most of my second-year PhD student friends still say that they don’t fully know what problem they’re actually trying to solve. Anyway, I do have an actual problem formulation that I will base my work on, and as long as I don’t get too caught up in side-quests like supervision and paper co-authoring I am poised for steady progress on my subject of choice. This makes me feel quite satisfied with how the first half year went for me and I’m optimistic for the next few months despite the initial excitement having worn down.
From my first experiences I can already report a few differences in the daily life of a PhD student that is employed by a company as opposed to a PhD student working for his/her professor alone (I will call these university PhD students from now on). Some things look better for me, some things look better for the university PhD students, and overall I’d say that both sides enjoy a similar level of comfort1:
1. My salary is slightly bigger than what I’d get as a university PhD student with a full contract (and thus much bigger than that of all the unfortunate PhD students that don’t get full contracts), but since university students in Germany get a raise after 1 and 3 years respectively my advantage really only applies to the first year.
2. Effectively my time as a PhD student is much more limited than that of a university PhD student. PhD contracts by default are time-limited, but a professor normally has access to multiple sources of funding such as research projects, government money etc., and can thus re-assign PhD students to different funding pools for continued employment. This means that the 6 years given by my university to finish the dissertation often is the actual time limit for university PhD students, provided their professor doesn’t want to get rid of them.
3. On the other hand most university PhD students will have to devote a large portion of their working hours to teaching/supervision or research project duties, meaning that they will often take longer than the intended 3 years anyway. As an industry PhD student I do not have any duties in my professor’s research group but I occasionally supervise students in Bachelor/Master theses and research seminars if the topic is interesting and fits my overall research bubble.
4. It probably depends greatly on where exactly you are doing your industry PhD, but I am very fortunate that I don’t have to participate in my team’s software development work (this might be different for your company). Don’t get me wrong, my colleagues at work are really cool and they are doing great stuff – but helping them out would take away portions of my limited research time, and so I limit my activity inside the team to keeping in touch, setting up calls if i’ve got questions and keeping track of their progress and overall vision.
5. Getting the tools and resources you need to do your research can be somewhat more complicated in an industry PhD because you’re bridging two worlds: Neither at my uni nor at my company do I have a dedicated office seat, and thus I work in 4 different locations including home office and another office of the company. Likewise, acquisition of compute resources and data for machine learning experiments is sometimes tricky because my university is not responsible for me and my company doesn’t see me as a regular employee.
6. Last but not least, living in two worlds makes it easy to meet many more people you can discuss and collaborate with – if I’m not careful I can easily drown myself in inspiring side-projects. However you’ll also have obligations in two communities and will thus have to figure out yourself how to cater to both of them. This especially applies to your advisor pair in academia and industry (given that they’re not the same person), where you’ll have to find a research topic that meets the expectations of both of them.
All in all, I am pleasantly surprised by the comfort of my industry PhD position and I think I’m in a very comfortable spot. Nevertheless it’s clear that time will run out quickly for me, so I should get going. It’s quite challenging to converge on a topic if you want to please two supervisors with very different backgrounds, but it’s also very fun and I’m still up to the challenge and optimistic about the outcome.
- Please keep in mind: I’m writing this from a computer science perspective – things might be different in your area.