Education
Tutorship for MSc Students
In general, each MSc D-MTEC Master student is required to select a professor from the D-MTEC as his/her/its tutor, ideally in the second semester. This tutor is responsible for approving the student’s supplementary course selection of the learning agreement. However, the supervision of the master thesis is not mandatory for the tutor. We refer to the MSc MTEC Regulations for details on the tutorship.
Please start by visiting our website to learn more about our research and, in particular, our labs. In general, we are looking for students with a strong technical background, especially in the fields of computer science and/or data analytics (data mining/machine learning).
For your application documents, please provide the following information in advance:
- Cover letter including a short motivational statement and your preferred lab,
- CV, and
- proposition of only supplementary courses (e.g., courses that begin with 363-XXX are not allowed since they refer to your main courses).
Please send your application documents to . Once we receive your application, we will forward your application to your preferred lab. Please note that incomplete or formally incorrect applications will not be reviewed. If we accept the tutorship, we will notify you by email.
As stated in the regulations, a tutorship is not offered for MAS students at D-MTEC.
We are looking for students with a strong technical background, especially in the fields of computer science and/or data analytics (data mining/machine learning). We do not require any mandatory courses. However, we strongly recommend taking one or more of the courses offered by our group, or courses related to the field of data analytics (e.g. 252-0220-00 Introduction to Machine Learning, 252-0535-00 Advanced Machine Learning, etc.).
If you are interested in writing your thesis with us, we would be happy to receive your applica- tion, consisting of a CV, a transcript of records, and a motivation statement outlining your topic of interest and identifying the potential host. For a list of possible topics, please check our webpage and contact the responsible person directly.
Semester and Thesis Projects
Scaling Expert Assessments Through AI and LLM-Based Reasoning: A Real-World Heat Pump Performance Assessment
system is configured correctly and running efficiently. This manual evaluation requires specialized knowledge, does not scale well, and can delay the detection of installation issues. With the rapid advancement of large language models and autonomous agent systems, there is a promising opportunity to automate parts of this expert reasoning process.
In this thesis, you will investigate whether LLM-based agents can understand and interpret heat pump performance data in a way that mirrors human expert evaluations. You will study how experts identify typical installation problems, derive the key features and patterns from time-series data that are necessary for assessment, and design an AI workflow capable of generating explanations and recommendations. The work includes preparing data, experimenting with different model architectures and representations, and building a prototype system that evaluates performance and flags potential issues.
The goal is to deliver a validated proof-of-concept that demonstrates how AI can support or partially automate expert feedback, ultimately enabling scalable, consistent, and timely assessments for large fleets of connected heat pumps. This project combines applied machine learning, energy systems knowledge, and agentic AI design, in collaboration with the ETH Agentic Systems Lab and the Bosch Lab.
Show details
AI for Supporting a Scalable Transition to Sustainable Heating in Residential Buildings
Heating systems in European homes were rarely designed with modern heat pumps in mind. When upgrading such buildings, one of the most important questions is whether the existing radiators can deliver enough heat to keep rooms warm under winter design conditions. Knowing the heating power of each radiator is essential for correctly sizing a heat pump, ensuring comfort, and avoiding costly mistakes. Yet most homeowners and many installers do not know which radiators are installed, how large they are, or what their thermal output is. Traditional assessments require manual measurements, catalog lookup, or accessing technical data that may be unavailable or difficult to interpret.
With advances in artificial intelligence - especially computer vision and reasoning models - we now have the opportunity to automate this entire process. Instead of manual work, a homeowner or installer could simply take a few photos of each radiator. An AI system could then recognize the radiator type, estimate its dimensions, and calculate its expected heating power at typical operating conditions. Such a tool would significantly simplify and speed up heat load calculations, which form the foundation for designing efficient, reliable heat pump systems in existing homes. This research aims to explore how AI can bridge this gap by interpreting real-world images and transforming them into actionable engineering information.
Show details
Bachelor/Master Thesis on Agentic AI in Education: Agents for Oral Student Examinations (ETH Zurich – Agentic Systems Lab, Lab Track)
This thesis project explores the use of Agentic AI systems in education, focusing on conducting oral student examinations with agentic AI. Students will work on an examination platform developed at ETH Zurich’s Agentic Systems Lab, which allows professors to upload course materials and automatically generate oral exam questions. The AI agent conducts examinations, asks follow-up questions, and evaluates responses in real-time. The thesis offers opportunities to contribute to the next generation of intelligent assessment systems by improving question generation, interactive reasoning, and evaluation accuracy, while investigating pedagogical and technological implications for higher education.
Show details