Mannheim Master in Data Science Erfahrung
Reflecting on my MMDS experienceTLDR: Great if you want to get into AI/ML/DS, even if you don't have a coding background.
Personal background
I went through the MMDS years ago (pre-ChatGPT). I applied with a BSc in economics and no traditional coding experience except some R and Stata.
Admission requirements
Minimum prerequisites are having (i) a bachelor's degree in a STEM/quantitative subject with competitive grades and (ii) an official proof of English proficiency. Non-EU citzens with a non-German degree also need to show decent GRE scores. Coding skills are not explicitly required.
General experience
A MMDS cohort ususally consists of only around 10-20 students. In my cohort half the students were German and the other half international. Profs are knowledgeable as well as approachable. All MMDS courses are taught in English. Many lectures are also attended by business informatics master students. The focus will be on fundamental theory, whereas coding skills need to be honed during exercise sessions with TAs, group projects and your free time.
Recommended courses
| course I liked | comment |
|---|---|
| Database Technology | Take it, if you haven't touched databases/sql before. |
| Large-Scale Data Management | Optimize sql queries in the context of distributed hardware. Probably useful for data engineering at Big Tech. |
| Data Mining I/II | Python and must-know data science methods. |
| Machine Learning | PyTorch and must-know ML methods. |
| Deep Learning | Methods like Transformers, LLMs etc. |
| Text Analytics | From NLP basics to language models. |
| Information Retrieval and Web Search | Handle web data and must-know retrieval methods. |
Career paths
The MMDS can prepare you not only for a career as a data scientist or ML engineer, but also a PhD, depending on your course selection.
Conclusion
All in all, I enjoyed my MMDS time. Especially, the introductory courses, which required group work (e.g., DM1), were particularly useful, since you can get your Python skills from zero to one by learning from/with your peers (although this has become almost trivial with the advent of LLMs). I still find myself using many of the taught methods and concepts in my day-to-day work.