Duration and Scope
Before completing the master program, you will have to do two lab rotations, each worth 10 Credits and each comprising the workload of 6 weeks of full time work. You can combine the two lab rotations into a single one comprising the workload of 12 weeks of full time work. You can split the total time of your lab rotation in other ways, e.g. do a single lab rotation in 12 weeks of half time work, or 18 weeks of a third time.
Before starting a lab rotation, a lecturer of the AI master program has to agree to be your formal supervisor for this. This lecturer does not have to actually supervise you (usually this will be done by your host), but will listen to your final presentation and, if all went well, give you the credits. Any lecturer of the AI master program can do this - you can browse the module handbook to see who is available.
Subject of your work
You are free to work on a lot of things. Ideally, in a lab rotation you should work on real life data, and understand it, clean it, then work on it, e.g. analyze it or fit a model that reflects the data well and could possibly be used for predictions. It should not be your task to gather the data yourself! And it should not be your sole task to clean up some large data base, or enter new data into it, but it can be part. You could also work on improving an existing Machine Learning method, testing some changes in the basic setup, or try out an entirely new Machine Learning method on toy data.
These are just rough guidelines - the lab rotation projects so far have been very diverse, and in doubt just contact a potential formal supervisor and ask if your envisioned project would be ok.
Your project does not need to lead to a successful application - failure is part of research.
Where to do the lab rotation
It is up to you. You can approach a research group at HHU, a research group at any other research institution, or a company. The group can consist of one person only as well. Lab rotations universities are typically unpaid, while lab rotations at companies are typically paid. If you already are working in a Data Science related job, it can be counted as a lab rotation, but ideally you should not just be doing one continuous task (like administrating customer data) but rather work towards a goal so that you have something to report on in the end.
For the start of the program in 2019/20 we assembled a small list of places that were willing to receive students for lab rotations. You can still ask at those places. There are many more companies based at Düsseldorf, and data scientists are in high demand, so you should find many more opportunities. Also, your lab rotation can take place anywhere in the world!
How to get your credits
At the end of the lab rotation you should give a talk (typically 30 minutes) to the group that hosted you, summarizing what you did. Invite your formal supervisor. If there is no venue for such a talk at your host group, talk to your formal supervisor - it won't be a problem. If your formal supervisor has no time to attend, someone from your host group can confirm what you did, and it should also be fine. After one of these things happened, your formal supervisor will give you the credits and they will appear on your students record. If your formal supervisor doesn't know how to do that, have them ask me (Peter).
Is there any paperwork around lab rotations?
Not from the side of HHU. Often companies want some confirmation that you are doing an obligatory internship as part of your studies. The reason is that in this case they don't have to pay a health insurance for you, while for voluntary internships they need to do this. Here are such confirmations, an English one and a German one. If your company wants something else confirmed, contact me (Peter).
Does the lab rotation need to lead to a thesis topic?
No. If it does, great, then you made a good investment of time and you will be deeper into your topic when you start working on your thesis. But it is not necessary.
What are the pros and cons for doing a single lab rotation versus two lab rotations?
Just the obvious ones: With two lab rotations you get around more and see different things. With one long lab rotation you get deeper into one topic. But see also the next point:
Can I do more than two lab rotations?
You can do whatever you want, but you will only get 10 credits twice (or 20 credits once, for a long lab rotation). There is one issue, though: Some companies will only take you when you are doing an obligatory lab rotation (see the paperwork section). After you got your 20 credits, a further lab rotation would probably not count as obligatory anymore. You are, however, free to do two long lab rotations and only have them count for 10 credits each.
What did previous students do?
Lots of different things, from lab rotations with lecturers of the AI program to internships in other countries. The research labs at Jülich have been a popular place, and there are a bunch of non-university research institutions in Germany, like the Max-Planck institutes and the Fraunhofer institutes.
There were attempts to start a private database maintained by you students, where people can tell about their lab rotations, but that didn't take off. You should give it another try though!