“What do I need to do, to be a successful Ph.D. student / How do I know if I was successful during my PhD project?” is a question I am often being asked. Hence, I decided to write some advice for (prospective) PhD students what they should achieve during their PhD time to be successful after their PhD when applying for postdoc positions (and professorships subsequently), research positions in companies or management positions. This advice relates particularly to doing a PhD in the field of machine learning, recommender systems, and information retrieval. Other fields may differ.
You will have lots of freedom during your PhD time, and you must prove that you are capable of making the most out of it. Someday, when you are a professor, nobody will tell you what to do. You are the one that needs to identify promising research problems, you are the one who needs to have ideas on how to solve them, you are the one who needs to acquire funding, and you are the one who needs to design lectures and supervise students. Nobody will take you by the hand. And it’s the same if you are a researcher in a company or a manager. The company hires you, and then you must perform with very little guidance from others. And “perform” means first and foremost one thing: Identify problems, prioritize them, and solve them (more precisely, you need to have ideas on how to solve the problems — the actual solving may be done by others).
Your PhD is your chance to prove that you can do this with little guidance from others.
You can consider the following as a checklist that will be used by others to assess whether you are a good independent researcher, teacher, and manager. Hopefully, by the end of your PhD, you can tick as many boxes as possible.
The most important points:
Good publications are the single-most-important metric for how good you are as a researcher. ‘Good’ publications mean the right mix of Top Quality (Top Conferences / Journals), High Quantity, and a good mix of first-author publications (with and without your PhD supervisor); last-author publications (you as supervisor of e.g. a Master student), and medium-author publications.
After that there comes for a long time… nothing. If I assess applications for postdoctoral positions, my first look goes to their publication list, and if that list is not convincing, I do not spend more time with that application. A lack of good publication cannot be compensated by anything. But everything else (see below) can be compensated by excellent publications.
Then, the following points are important, in no particular order (keep in mind that you do not need to achieve all of them, but achieving at least some would be beneficial).
- Acquired Funding (e.g. a conference travel award; a major contribution to a ‘real’ grant would be excellent)
- International Experience (Research Visits abroad, ideally in a country that speaks not your native language; demonstrate that you are mobile and flexible. Worst Case: Bachelor, Master and PhD in the same place without any stays abroad).
- Your work had an impact in the academic community (demonstrated by citations of your work)
- Your work had some real impact (e.g. is used by companies; you did a business start-up yourself; your work is public on e.g. GitHub and has many forks/stars; many followers on Twitter; some newspapers reported about your work)
- Demonstrate management skills (through successful student supervision — “successful” as in the output lead to e.g. a publication, some software on GitHub or at least a nice blog post)
- Awards (Best Paper; Best Reviewer; Scholarships; Business Plan Contest; …)
- Large Network of collaborators (expressed e.g. through co-authors)
- Engagement in the community (Reviewer for conferences; Support in organizing conferences or workshops)
- Teaching experience (and ideally positive evaluations)
- Industry Experience (Can also be pre- or post PhD)
- Invited Talks (this would be an exception for PhD students, but excellent if you achieve this)
- Affiliation with prestigious institutions and authorities in the field. This can be achieved, obviously, by doing a Ph.D. at a prestigious institution but also and additionally by research visits, invited talks, and guest lectures at prestigious institutions as well as through joint publications with authorities in your research field.
- Passion for some non-tech stuff (demonstrate that you have a life beyond doing research)
The next important question is, how can you achieve all these things? The answer is simple. You need:
- A PhD topic that you love
- Be the right person for a PhD
- Solution-oriented and a natural-born problem solver
- Extremely self-motivated
- Resistant to disappointment
- Self-Reflective and open to criticism
- Intelligent (of course, the more the better but, frankly, ‘average intelligent’ is good enough if the other criteria are met)
— long time nothing —
- A supportive environment
- A good supervisor (good skills, good network, enough time for you, truly interested in your success)
- Smart PhD students and PostDocs in your group
- A good reputation of your host-university (this can open doors)
- A country in which it is easy to acquire funding e.g. for research visits, conference travel, or business start-ups
- Good office space
- A good city to live
- Flexibility to adjust your PhD topic when needed (important: it’s always ‘when needed’ not ‘if needed’). This can be difficult particularly in industry-related PhDs.
- Enough time/flexibility to do all the above-mentioned points (there can be huge differences in administrative workload or ‘side’ projects among different universities and funding schemes)
- Keep as much of your IP as possible. This is actually an important point that many PhD students do not think about. Depending on the funding and the institution you may or may not keep (parts) of the IP you generate. For instance, you might create some great software during your PhD — what if you want to do a start-up afterward? In the worst case, your university or sponsor owns everything and does not allow you to use your source code. Policies on IP may also prevent you from releasing your work as open source. Hence, before starting a PhD ask your institution
- What exactly of your output is considered IP (source code? ideas? websites? …?)
- What can you do with the IP? Can you publish source code and data? Can you publish all your research results, or could someone (e.g. industry sponsor) prevent you from publishing certain parts? Would you be allowed to use e.g. source code beyond your PhD for your Postdoc (at a different institution) or a business start-up?
- Decent Funding (e.g. for equipment and conference travel)
- Cumulative dissertation (this is a huge plus that allows much more flexibility in your PhD topic, saves you from many demotivational moments, and saves you, at least, several months of work. Maybe even more importantly, it allows you to collaborative more closely with fellow PhD students in your group.