Both for AutoLab-Next Automation GmbH and for the DigiThy project, I have already had the opportunity to supervise and support several bachelor's and master's theses.
This repeatedly raises the question: When is it actually worthwhile for a company or a research project to offer students a thesis?

Because even if at first glance it seems like a simple win-win situation, especially due to the manageable costs, the supervising company or institute still faces a not insignificant amount of effort. Topics must be appropriately chosen, clearly formulated, and embedded in such a way that they provide real added value for both sides: students and clients.

Many considerations about when it makes sense to announce a new thesis are similar, regardless of whether it is a purely academic thesis or a more practice-oriented one involving a company. However, there are differences in some aspects.

Based on my previous experiences, I would like to show when theses are truly meaningful in practice, what differences exist between decisions in companies and in academic research projects, and what should be considered in order to derive the greatest possible benefit from them.

Personal experiences from practice

As part of the DigiThy project, in which I am involved at the Institute of Control Engineering at TU Graz, I have had the opportunity to supervise and support several master's and bachelor's theses. One early master's thesis was based on a model I had previously developed for Graves' disease and focused on model-based approaches to treating the condition. Later, theses followed in the field of reinforcement learning. In addition, I supported several bachelor's theses, including the development of a web-based treatment interface and the use of smartwatches. Depending on the project, my level of involvement varied, but in every case I was able to gain valuable insights into the supervision of scientific work in research.

For our sister company Promotech, I had the opportunity to support a bachelor's thesis in which we dealt with complex motion controls. A central component was the meaningful use of unit tests to ensure clean implementation. I’ve already explored in detail why proper testing and structured software development should be integral parts of scientific practice in my first blog post.

At AutoLab, a bachelor's thesis achieved the execution of AI network inference for assessing the quality of image data directly within the real-time capable part of a PLC control system. Additionally, a master's thesis on digital twins for predictive maintenance was completed, in which several of my favorite topics—namely physically motivated modeling, data-driven decision-making, and optimization—could be combined. Another master's thesis applied and further developed algorithms for pick-and-place applications based on 3D image data. Here too, AutoLab significantly expanded its expertise in the field of image processing. Through these activities, I was able to gain valuable experience in how work at the intersection of research and industry is carried out.

Basically, each of these theses had added value, even if it was expressed to varying degrees. The results of the pick-and-place applications were somewhat overlooked in AutoLab’s day-to-day operations, and their full potential was not realized. As I also described in another blog post, it is often the case with academic theses that after the handover, a large part of the work that has already been done is not immediately usable, resulting in a significant loss of time for re-familiarization. Frequently, the codebase at the end of a thesis is not optimally documented or structured, which leads to valuable knowledge being lost from one master’s thesis to the next.

To learn from this, I would like to show below what prerequisites must be created so that master's theses can reach their full potential and provide real added value for both companies and research groups.

Prerequisites for getting the most out of scientific theses

Choice of topic

In my experience, there are clear differences in the choice of topic between purely academic theses and theses initiated by companies.
The scientific context allows for an in-depth engagement with open questions without the need for an immediate benefit or direct application to be apparent. Topics often arise from reading something interesting and wanting to pursue new approaches based on that. This openness can lead to significant research progress in the long term, even if the immediate value is not yet clearly visible at the beginning.

The situation is different for theses that arise in a corporate context. Here, an open-ended idea is not enough. Already at the stage of topic selection, it should be clear what specific added value can be created for the company. This does not mean that the thesis must directly result in a finished product that can immediately generate revenue. But at the very least, there should be a visible possibility of later deriving a product, a new process, or a valuable internal solution from it. If this perspective cannot even be outlined, the thesis usually turns out to be a complete waste of time for the company.

Clear goal definition and expectations

The question of goal definition is directly tied to the choice of topic. Here too, I see clear differences between academic theses and those in a corporate context.
In the purely academic field, the definition of goals can be kept somewhat more open. Students should have the freedom to try out new approaches, and this openness is often a prerequisite for truly innovative ideas to emerge. Good scientific work thrives on not limiting oneself too early.

In company-initiated theses, however, the definition of goals must be clearer and more strictly defined. Openness remains important here too, because even in a corporate environment, students must have the opportunity to develop their own solution approaches. But the framework within which this openness takes place should be more narrowly set.

Another important point concerns the type of documentation. For the academic degree alone, the final scientific thesis may be sufficient. For me personally, however, that is never enough. I consider an additional, practice-oriented documentation of the tools, frameworks, and code components used to be absolutely necessary in every case. The criteria for this should also be defined in advance. Why this is so important and what form this documentation can take will be explained in the following sections.

Good onboarding and knowledge transfer

Especially in follow-up projects that build on existing topics, a well-maintained knowledge base is invaluable for onboarding. Clean documentation of all relevant information enables new team members or students to become productive much faster and to build on previously developed results – instead of having to start from scratch again.

If such a documentation base does not yet exist, the start of a new project should be used as an opportunity to systematically gather and record the existing knowledge. At the beginning of the thesis, it is primarily the supervisor's responsibility to structure and summarize the information. But this shouldn't turn into a doctoral dissertation. What matters most is to write down the key information briefly and clearly. Especially when documenting the initial information, the rule is: you don’t have to know everything right away. The goal is to start the work in a meaningful way. Open questions or uncertainties are not only allowed but should be noted. What’s important is that they are documented at all, because that provides orientation and makes further work easier. As a tool for documentation, I recommend Confluence. If you’re interested in my favorite project tools, I suggest reading one of my previous blog posts.

Another key component is a joint kick-off meeting attended by all parties involved, including supervisors, students, and professors. This setting allows for the discussion of organizational matters, technical requirements, and initial scientific questions. This ensures that everyone has the same level of information and that the thesis begins on a solid foundation.

What I personally now consider at least equally important is getting to know each other on a personal level at the beginning. A bit of team-building often works wonders, whether it’s over a shared lunch, a coffee, or a casual conversation away from the technical topics. Just recently, I had the opportunity to supervise a master’s project in the field of reinforcement learning, which built on an existing master’s thesis and on the DigiThy framework in general. To get everyone on board, I invited the professor, the previous student, and the new student out for burgers. That really brought the group together and created a relaxed, open working atmosphere.

Supervision with regular feedback loops

After a successful kick-off, it is important to carry the momentum into the actual working phase. What has proven effective in practice is a regular meeting approximately every two weeks.

This meeting can serve as a point of orientation, but it should not be handled too dogmatically. Especially during stressful periods such as exam times, it makes sense not to insist on rigid procedures but to respond flexibly to the actual needs of the students. Quieter periods, in turn, can be used more intensively. When researching something new, it is always a creative process. And creativity cannot be forced. That is why it is important to maintain sufficient flexibility, both in a corporate and in a purely academic environment. Clear structures can be helpful, but they should not be so rigid that they hinder initiative and new ideas.

What has worked well for me is uncomplicated and always-available communication. I usually give students my number directly and tell them they can reach out via WhatsApp at any time. Especially students in the field of computer science often have rather irregular or chaotic working hours. Many work in the evening, at night, or on weekends. In such cases, it's helpful to be able to quickly ask a question in the middle of their workflow without having to wait for the next official appointment. Of course, this is a matter of personal preference, but it works well for me. I genuinely enjoy supervising theses and have no problem being available at all times.

Focus on documentation and sustainability

In my opinion, good and comprehensive documentation is essential in both areas — academic and corporate. In the academic field, the scientific thesis is often considered the main form of documentation, which is generally sufficient to present the subject matter in a comprehensible way. Nevertheless, care should also be taken to ensure that developed methods, tools, or code are cleanly documented in addition, in order to facilitate future continuation of the work.
In the corporate context, practice-oriented documentation such as a knowledge base, a wiki, or structured technical documents is often even more important, as continuity and direct usefulness are more strongly emphasized. Only with clean documentation is it possible to make results sustainably usable and to significantly reduce onboarding times between consecutive theses or projects.

In general, it is standard practice for theses to be presented at the institute. If scheduling allows, supervisors should definitely attend these presentations.

Additionally, I consider it very useful for theses with a corporate connection to also be presented within the company itself. A presentation to the relevant teams ensures that the results become visible and often opens up new follow-up opportunities. Especially in the case of master's or bachelor's theses with practical relevance, this is a good opportunity to share knowledge internally and to spark discussions that go beyond the specific thesis.

When are scientific theses worthwhile?

For companies

In company-related theses, the question of cost plays an important role. Even if students are not paid directly, internal efforts arise, such as supervision by experienced employees with corresponding hourly rates. As a result, several thousand euros in personnel costs can quickly accumulate from internal resources alone. That’s why careful consideration should be given in advance as to whether the topic can provide real long-term value for the company.

Additional costs often include compensation for the students as well as a supervision fee for the institute. Altogether, this quickly adds up to a considerable effort. The benefit of a scientific thesis is rarely immediate. Even with a successful outcome, the path from prototype to finished product is often significantly longer and more expensive, in terms of both time and budget. That is why it should already be clear before the topic is announced whether it can be economically viable throughout the entire process up to its practical application.

For research institutions

Internal personnel costs also arise with scientific theses, yet the decision does not have to be based solely on economic considerations. Often, bachelor's or master's theses contribute to a larger research project and make a meaningful contribution to the overall effort. What matters is whether the supervised topic generates usable scientific output, such as papers, frameworks, or tools for follow-up projects and funding proposals. Therefore, the ratio of time investment to scientific benefit should be clearly assessed in advance.

If I had known that earlier

The thoughts mentioned above can already be understood as a small guide. In this section, I would like to summarize what I have personally learned from supervising scientific theses so far.

First: There must be a clear added value. In the corporate context, this means asking whether a product, a solution, or at least a usable prototype can result from it. In the academic field, it is about whether the work can lead to publications or further research.

Second: Supervision takes time. I once took on a master's thesis while AutoLab was running its largest project to date. In hindsight, that was a mistake. The supervision fell short, and the results were barely usable for me. Since then, it has been clear to me: only supervise if you truly have the capacity to do so.

And third: My personal evergreen – work cleanly. That means above all testing, documenting, and testing again. Those who are thorough in this regard can not only make better use of the results but also lay the foundation for truly strong projects.

Personal conclusion and outlook

What I can recommend to everyone: supervising scientific theses is simply a lot of fun. You learn a great deal yourself, gain new perspectives, and work with people who are motivated, curious, and enthusiastic. That’s exactly what makes this collaboration so rewarding – not just for the students, but also for me as a supervisor.

You come across ideas and solutions that you would never have come up with on your own. At the same time, such theses bring fresh momentum into the company. They break through entrenched ways of thinking and encourage a more open and creative mindset. Especially in a corporate environment, where processes and routines often dominate, this is incredibly valuable.

My personal outlook is clear: I want to remain active in both areas. With AutoLab, I want to continue working at the forefront of technological developments and driving innovation forward. At the same time, I am committed to helping shape the academic field – especially where research brings real medical benefits. I look forward to the topics, projects, and people that will accompany me on this journey.