How should I start my own private post-doc research project?

Question

I am interested in pursuing my own part-time postdoctoral research in computer science, with a view to getting one or two publications over the next two to three years. My motivations are personal interest, and also to keep my options open in case I decide to apply for an academic or research post in 2016.

I am not sure where I should start looking for a project. Should I read academic papers until I get ideas? Should I just try and brainstorm my own ideas? Should I go back to my current university supervisor, or my previous PhD supervisor?

My Background

I am nearly 41 and married with two children (aged 10 and 13). I gained my bachelor's in computing in 2006. In 2012 I submitted a thesis in artificial intelligence (case-based reasoning) and the PhD was awarded in 2013. My family are not keen on moving and we live in an area with very few high-tech or academic employment opportunities.

I can't see a way to extend my PhD research. I have never worked as a post-doc; before the PhD I was a database administrator and immediately afterwards I was a software developer.

Fortunately, late in 2013 I found a position on a UK government scheme known as Knowledge Transfer Partnerships. I am employed by a Scottish university to work within an English manufacturing company, to use AI techniques to solve a particular engineering problem. I am based at the manufacturing company and I see my university supervisor very rarely. I am not allowed to publish anything from this employment. The contract ends in August 2016.

The KTP project will probably use standard AI methods; I won't be inventing any new techniques. If the manufacturing company consented (and they will not!) it would be possible to publish, but it would only be as an interesting application of AI, not a new way of doing AI.

I have a substantial training budget (GBP £2000 a year) which—with the approval of my employer—I can spend going to conferences, buying books etc.

I have discussed the idea of doing research with my university supervisor, either performing experiments myself, or a literature review paper. She made some nice noises but she was not overly enthusiastic. I think she is worried that it will divert me from the KTP project (for which her university gets paid a lot of consultancy money) and possibly upset the construction company who I am based with.

My interests and skills lie in artificial intelligence (neural networks, case-based reasoning), computer-aided design and software engineering.

I still have my PhD supervisor on my LinkedIn page and I could approach her. However, I worry that she will think it is strange if I ask to collaborate with her, given that I am currently employed by a different university.

I can commute to my current workplace (it's about an hour away) and there are 2 good universities (and another two not-so-good ones) about 60-90 minutes' commute from our family home, so location is the biggest restriction for me, I can't force my wife and children to move.

Answer 1

Partial answer: start reading!

Read for both breadth and depth, and be sure to occasionally step outside the boundaries of what you are familiar with, even exploring fields that seem only tangentially related. In the process, you will almost certainly discover an area that is begging to be explored further. Indulge yourself, dive into the literature, buy books and attend conferences (for which you say you have a budget), and keep excellent notes while doing so.

You may discover that you can write a literature review fairly easily, and will probably also find that you have identified exactly what you wish to pursue for your research project. At worst, you will have spent many enjoyable hours increasing your knowledge; at best, you will be well on the way to completing your research.

Answer 2

Doing "unsponsored" research would be tricky, but at least it's more feasible in artificial intelligence than in other fields (such as experimental particle physics or cell biology).

In general, however, working with your PhD advisor on new projects after you have finished your doctoral research is considered a very bad thing. The reason is that this suggests that you are still dependent on your graduate advisor, and are effectively still "riding her coattails." Consequently, it suggest you're not ready to stand on your own, which makes you much less desirable to hiring committees. Furthermore, in places like Germany, if you have only published papers with your graduate advisor, one can argue that you're not actually yet qualified to be an independent principal investigator, which can reduce your ability to apply for and receive grants.

Note that this does not mean that you can't finish up papers that are part of your graduate thesis work after you graduate. It just means that you shouldn't start up anything new.

Answer 3

What do you want to achieve by doing research "on the side"? Is it purely for intellectual stimulation? Do you want to maintain/raise your profile in the field? Are you hoping to create something that leads directly to your next job (through a research grant or similar)? Thinking about these questions may help you work out where to go.

First, you need to consider whether your current (KTP) employer is likely to raise any conflict of interest issues, and ensure you steer clear of any potential difficulties. Then, think about your PhD work: are there any outstanding issues that you could investigate? Projects that never got written up? In your situation, anything where you have a head start on the research is valuable. I'd definitely get back in contact with your original supervisor - they may have ideas, or contacts, or simply advice. Whilst it's true that in an ideal world, you'd expand your pool of collaborators, anything is better than nothing.

Are there any research groups active in your field in the local universities? Make contact, find an excuse to visit - it can't hurt. Even better, are there any researchers from other fields attempting to apply AI to their own problems, or trying to solve problems that are suited to AI? (Almost certainly!) By bringing your knowledge to another field, you may be able to achieve much greater impact for a given amount of effort! And if you offer something that most people in that field don't have, you'll find people are much more keen to collaborate. If you make yourself sufficiently useful to a large and well-funded research group, you may find a job offer down the line. So, spend some time investigating what your local universities are good at, find something that interests you and you think you can contribute to, and make contact. They might not be interested, but it doesn't hurt to try (and you may find postdocs/junior faculty more receptive to developing a side-project than senior professors).

I guess the bottom line is: be pragmatic. Unless you're only interested in the intellectual stimulation, in your present situation, you need to focus on maximising the cost:benefit ratio. This might mean not working on the things that most interest you, for the time being.

Good luck!

Answer 4

The best way to start a project that will get funded is to read funding announcements (FOA, PA, RFA, RFP, etc.) in your areas of expertise and/or areas that really interest you. Funding announcements are availible directly from the funding agency, or through online database such as Community of Science and Grants.gov. Reading funding announcements is an easy way to see the current state of knowledge as well as the gaps that exist (and available funding!).

I disagree with some of the other comments, collaborating with your PhD supervisor is very common (in my field). Furthermore, cross institutional collaboration is encouraged as pointed out by Luke M. I see no downside to this.