Markus Frohmann
NLP Researcher
- I started with a Bachelor's degree in the AI program at JKU Linz in Austria, led by Prof. Sepp Hochreiter, the inventor of the LSTM. On the side, I started to work on making AI more accessible to customers, first at a startup, and later at the German company Bosch. I enjoyed it; however, I then started working on my bachelor's thesis - and have loved doing research since then! I worked on multiple projects in international settings and have published at and attended multiple conferences. My current interests involve Natural Language Processing and Multimodal AI.
I’ve worked extensively with high school, undergraduate, and doctoral students—helping them explore interdisciplinary topics, develop research skills, and build confidence as independent thinkers. I’m excited to share my experience to support the next generation of curious, driven learners.
- Universities attended -
- National Taiwan University of Science and Technology National Taiwan University of Science and Technology MSc (Exchange Semester within Master's), Computer Science
- Johannes Kepler Universität Linz Johannes Kepler Universität Linz Master of Science - MS, Artificial Intelligence
- Johannes Kepler Universität Linz Johannes Kepler Universität Linz Bachelor of Science - BS, https://www.jku.at/studium/studierende/akademische-feiern/
- AHS Köflach AHS Köflach High School Diploma
- Preferred format of mentorship - Flexible / I’m open to different formats
- Time zone - Taipei (GMT+8); but can be very flexible
- Your general weekly availability for sessions - I am free most weekdays and can be flexible during them, preferrably from 9am-12am or 1pm-5pm (local time; evening can also be arranged9
- Any additional information you’d like to share with us? -
- Currently residing in the Savannah, US. Might change my location in few months but will keep you informed about anything in advance.
- Brief Summary of Your Research Interests.
- I want to make AI models more modular, with a focus on multimodality and multilinguality. Think of it this way: A model like ChatGPT is like a monolith: large, slow, and inconvenient to adapt. What if you want it to be better at your (non-English) language? What if you want it to be compatible with 3D scans? Currently, adapting models in this way is not possible, and I do not like this. So, I envision a future of model building like LEGO: It should be easy to add new parts (dialects, tools, ...), and everyone should be able to do so! For this, we need modular AI with an open and democratized ecosystem.
- Please describe your past experience mentoring or teaching students.
- I have been tutoring high school students for math and English for multiple years. Moreover, I am Head of Education (Brainery) department and Co-Founder of the first student-run initiative for AI, neuron.ai, in Austria. I am also mentoring a Master's student in his thesis at my university. Finally, I have led multiple research projects (our MICCAI Workshop papers and 2 currently in progress)
- List 3–5 example project ideas students could pursue with your guidance.
- Why do models like ChatGPT perform so well on English or French, but poorly on other languages such as Swahili or Hakka?
- What can be done to improve the quality of language models (like ChatGPT) in other languages? (data, training, ...)
- How can we adapt AI models easier and faster to new settings (languages, personalities, ...)?
- What can we do to make AI development more open and democratized?
- How can we make AI systems behave more independently, e.g., to browse the internet or solving a harder task?
- What types of final deliverables can your students expect to produce?
- Research paper
- Literature review
- Data analysis report
- Codebase or technical prototype
- Website or app
- Infographic or digital design
- Presentation or oral defense
- Any additional project themes, trending ideas, or real-world challenges you’d love to explore with motivated high school students?
- Bridging the Digital Divide in Education: Exploring access to technology and internet as a barrier to learning in underserved communities—and designing potential solutions. Mental Health and Academic Pressure in High-Achieving Students: Investigating the social and psychological impacts of competitive academic environments and proposing interventions. Sustainable Innovation in Engineering and Design: Examining how engineers can create environmentally responsible solutions to global challenges like waste, energy, or water access. Bias in Algorithms and Tech Design: Analyzing how social biases can become embedded in AI systems and exploring ways to promote fairness and accountability. Culturally Responsive Curriculum Design: Researching how school curricula can be made more inclusive of diverse histories, perspectives, and learning styles.
- 🗣️ Communication & Presentation
- Academic writing (papers, reports)
- Public speaking / oral presentations
- Creating research posters or infographics
- Storytelling through data
- Making a compelling final product (website, video, podcast, app)
- 🔍 Research & Inquiry Skills
- Formulating research questions
- Conducting literature reviews
- Identifying credible sources
- Understanding academic research structures (papers, abstracts, citations)
- 🧠 Critical Thinking & Analysis
- Analyzing arguments and data
- Identifying biases and limitations
- Evaluating conflicting evidence
- 📊 Quantitative & Qualitative Methods
- Designing experiments or surveys
- Data visualization
- Coding for data analysis (Python, R, etc.)
- 🎨 Creativity & Problem-Solving
- Connecting interdisciplinary ideas
- Applying knowledge to real-world challenges
- 📅 Project Management & Organization
- Setting research goals and timelines
- Documenting progress
- Revising based on feedback
- 🤝 Collaboration & Mentorship
- Working in small research teams
- Engaging in intellectual discussions
- Receiving and applying constructive feedback
- Building research confidence and autonomy