Student Projects
The DaST group is always interested to supervise student projects and theses that contribute to the on-going research in the group. If you are interested, please send the project choice from the following list and your CV by email to a group member.
Current Student Projects & Theses
- Using Cardinality Upper Bounds to Infer Variable Orderings for Worst-Case Optimal Join Algorithms (MSc project, February 2026 -)
Students: Andreea-Maria Chiriac, Daniel Tran, Zahraa Zaiour - Vision-based rock manipulation using a excavator with impact hammer (MSc thesis, February 2026 -)
with ETH Robotic Systems Lab
Student: Bingjie Xue - Digital Biomarker for Neurorehabilitation from IMU Longitudinal Dataset (MSc project, February 2026 -)
with ETH SCAI Lab and Lake Lucerne Institute Dart Lab
Students: Adnana Ivana, Doga Mentese, Tuna Ozdemir, Li Jiatong - Scaling Expert Assessments Through AI and LLM-Based Reasoning: A Real-World Heat Pump Performance (MSc thesis, January 2026 -)
with ETH Information Management and Bosch IoT Lab
Student: Guanyu Chen - Advancing (Sparse) Multi-View Depth Estimation (BSc thesis, January 2026-)
with ETH Computer Vision and Learning Group
Student: Fabio di Meo - Correction of errors generated by an OCR system employed on ventilator interfaces through post-OCR processing techniques (BSc thesis, January 2026-)
with Hamilton Medical
Student: Borislav Rakic - Automatic Analysis of The Marginal Bone Level Around Different Dental Implant Systems In 2D Imaging (MSc thesis, January 2026 -)
with UZH Clinic of Reconstructive Dentistry
Student: Nina Rubesa - Adaptive Conversational Support for Long-Term Robot-Assisted Rehabilitation (MSc thesis, December 2025 -)
with ETH Rehabilitation Engineering Lab
Student: Xuan Ji - Implementation and optimization of the PANDA algorithm (BSC thesis, December 2025 -)
Student: Florian Mattmüller - AI and ML Techniques for Short-Term Energy Price Forecasting (MSc thesis, November 2025 -)
with Axpo Group
Student: Rohit Koonireddy - Critical Questions for Text-to-SQL: Scheme-Guided Multi-Agent Reasoning with LLM (MSc thesis, October 2025 -)
with ZHAW Intelligent Information Systems Lab
Student: Xinyao Cao - Using Large Language Models to Represent and Forecast Temporal Dynamics in Sequential Data (MSc thesis, October 2025 -)
with the BMIC group @ ETH CVL Lab
Student: Tamanna Kumavat - A Retrieval-Augmented Generation System with Guaranteed Course Alignment for Data Science (independent study, October 2025 -)
Student: Giuseppe Doda - Diminished Reality for Personalized Recommendations to Support Healthy Food Choices in Multi-User Scenarios (MSc project, October 2025 -)
with Chair for Interaction- and Communication-based Systems, University of St. Gallen
Students: Ruihan Gao, Yiting Shang - Benchmarking Supervised and Self-Supervised Mortality Risk Prediction Models on ECG Data (MSc thesis, September 2025 -)
with UZH Quantitative Biomedicine
Student: Yanyang Gong - Enhancing Multi-view 3D Reconstruction: Diffusion-Augmented Gaussian Splatting Based on NoPoSplat (MSc project, July 2025 -)
with ETH Computer Vision and Geometry Group
Students: Chenhuan Cai, Yiming Xiao, Yuying Zhong
Completed Student Projects & Theses
- Automated 2D-3D image registration of kneww kinematics from fluoroscopic images using differentiable renderers (MSc thesis, July 2025 - January 2026)
with ETH Lab for Movement Biomechanics
Student: Qingyu Jiang - Should Human or AI Lead the Creative Process? (MSc thesis, July 2025 - January 2026)
with ETH Strategic Management and Innovation Chair
Student: Taishuo Zheng - Towards Data Efficient Predictive Maintenance for Industrial Pneumatic Components of Varying Types and Sizes (MSc thesis, July 2025 - January 2026)
with ZHAW/SMC Competence Center
Student: Weimin Yang - Symbolic Matrix Decompositions in Symbolica (MSc independent study, Nov 2024 - Nov 2025)
with Symbolica
Student: Zisen Liu - In-silico Directed Evolution using Protein Language Models with an Application to RNA Editors (MSc project, June 2025 - October 2025)
with Chair of Medical Informatics, Department of Quantitative Biomedicine, UZH
Students: Uros Dimitrijevic, Nuri Tas - Optimal Spike-based Neuromorphic Processing for Wearable EEG (MSc thesis, March 2025 - October 2025)
with Neuromorphic Cognitive Systems group, Institute of Neuroinformatics, UZH
Student: Alen Frey - Quantizing Vision and Language Foundation Models for Efficient Inference (MSc independent study, February 2025 - October 2025)
with ETH Integrated System Lab & ETH Computer Vision Lab
Student: Qingxuan Chen - Surgical Phase Recognition using Vision-Language Models (MSc thesis, March - August 2025)
with ETH Computer Vision Lab
Student: Junyong Cao - Leveraging Multi-Task Learning for Improved Spatial Transcriptomics Prediction from Hematoxylin and Eosin Images (MSc thesis, Feb - August 2025)
with Max Plank ETH Center for Learning Systems
Student: David Lebrec - Location specific collaborator suggestion based on papers (MSc thesis, Feb - August 2025)
with ETH Neuroinformatics
Student: Grigor Dochev - Output-Sensitive Evaluation of Conjunctive Regular Path Queries (BSc thesis, March 2025 -July 2025)
Student: Alexandru-Mihai Hurjui - Machine Learning-Guided Combinatorial Optimization for Cost-Effective Passenger Rebooking (MSc thesis, Dec 2024 - June 2025)
with SWISS Airlines
Student: Georgia Lazaridou - Randomized Linear Algebra over Database Joins (MSc study, Oct 2024 - March 2025)
Student: Deborah Stäubler - Automatic Analysis Of The Marginal Bone Level Around Dental Implants (MSc project, June 2024 - May 2025)
with UZH Clinic of Reconstructive Dentistry
Students: Xinyao Cao, Xuan Ji, Petrovic Teodora, Rubesa Nina, Xue Wang - Implementing factorized incremental view maintenance on top of Flink (MSc thesis, April - December 2024)
Student: Yizhi Zhang - Counterfactual prediction-based analysis of spatial transcriptomics (MSc thesis, April - Oct 2024)
with Center of Experimental Rheumatology (University Hospital Zurich)
Student: Ming Yi - Dynamic Model Counting (internship, Feb - July 2024)
Rémy Kimbrough (Ecole Normale Supérieure, Paris) - Tensor Decomposition by FAQs (MSc thesis, April - Oct 2014)
Student: Yuchen He - Query Language Support for the FRANTIC query engine (BSc thesis, March - Sept 2024 )
Student: Timo Lennard Tietje - Efficient Computation of Banzhaf and Shapley Value for Database Queries (MSc thesis, March - Oct 2024)
Student: Samuel Andreas Brügger - Improving Prompting Strategies for Reasoning Tasks (MSc thesis, Jan - July 2024)
with NLPED lab (ETH)
Student: Adarsh Shivam - Leveraging Artificial Intelligence to Enhance Education in High Schools (MSc thesis, Feb 2024 - July 2024)
with NLPED lab (ETH)
Student: Alessandro Vanzo - Wrist-worn Accelerometer-based Chronic Disease Classification: Model-driven and Data-Driven Approaches (MSc thesis, Nov 2023 - May 2024)
with Health IS Lab (ETH)
Student: Songyi Han - Efficient Query Maintenance using Maximal Hierarchical Subqueries (MSc thesis, Nov 2023 - May 2024)
Student: Neng Xu - Dynamic Computation with Static and Dynamic Relations (MSc thesis, Nov 2023 - April 2024)
Student: Zheng Luo - A Hand-Tailored Benchmark for Celltyping on ASAP and BGEE Resources (MSc project, April 2023 - March 2024)
with Swiss Institute of Bioinformatics
Students: Ming Yi, Liyuan Rong - Cardinality Estimation using Lp-norms on Degree Sequences (MSc thesis, Sept 2023 - March 2024)
Student: Luis Torrejón Machado - SemantIQ: Semantic Query Answering with Natural Language on Knowledge Graphs (MSc project, Oct 2023 - Feb 2024)
Student: Jan Luca Sheerer (ETH) - M-Flow: Incremental View Maintenance of Multiple Query Workloads under Updates (MSc thesis, Feb - Nov 2023)
Student: Rui Zhou (ETH) - Diverse Answers to Queries (MSc project, Jan - Nov 2023 )
Students: Jun Tu, Thi Phuong Anh Pham, Julian Novoa Martin - Applications of Transformer-Based NLP in Recruitment: A Use Case in Education (MSc project, March - Oct 2023)
with the UZH Chair for Development and Emerging Markets
Students: Alessandro Vanzo, Le Hoang Minh Trinh, Maria Korobeynikova - KroneDB: Factorized Databases meet Kronecker Products (MSc thesis, Feb - Oct 2023)
Student: Thomas Rolf Mannhart - CaVieR: Multiple Query Maintenance using Cascading View Trees(MSc thesis, Jan - Aug 2023)
Student: Johann Schwabe - Adaptive Factorised Data Representations via Reinforcement Learning (MSc thesis, Jan - Aug 2023)
Student: Christoph Mayer - Convex Hull for SVM(MSc project, April - July 2023)
Students: Pascal Severin Andermatt, Luis Torrejon Machado - Federated Learning for Respiratory Disease Classification from Audio Recordings in a Pandemic Scenario(MSc thesis, Jan - July 2022)
with Department of Management, Technology and Economics at ETH Zurich
Student: Mesut Ceylan -
- Approximate Convex Hulls (MSc project, July 2022 - Jan 2023)
- Students: Pascal Severin Andermatt, Tobias Fankhauser, Paul-Philipp Luley, Luis Torrejon Machado
-
- Matrix Profile (Independent study, March - Sept 2022)
- Student: Christoph Mayer
-
- Dynamic Query Evaluation(Independent study, Dec 2021 - June 2022)
- Student: Erlin Zylalaj
-
- Feature Engineering (MSc project, July 2021 - Jan 2022)
- Students: Carmen Christen, Lukas Vollenweider, Christoph Mayer
Strand 1: In-Database Machine Learning
Goal: Train machine learning (regression, classification) models over data matrices defined by relational queries. The project may focus more on theoretical or systems-building aspects, depending on the strength and interest of the student. The exact task (e.g., which ML task) can be agreed with the project supervisor.
Outcome: Good understanding of the research landscape on in-database machine learning; Design of a novel learning algorithm that exploits the relational structure of the underlying data to lower the computational complexity of the learning task; Implement the algorithm and benchmark it against existing solutions; Write-up a detailed report.
Context: Recent publications and talks (pdfs and videos) available at our publications page
Prerequisites:
- Good programming skills in C++ or Julia
- Conversant with: computational complexity; data structures and algorithms; databases (good grades in undergraduate courses on these topics); elements of machine learning (good grades in courses on Foundations of Data Science or Stats)
- Desire to push the frontier of knowledge at the interface of databases and machine learning
Strand 2: Linear Algebra over Relational Data
Goal: Understand fundamental linear algebra operations over matrices, where the matrices are defined by queries over relational data. The operations of interest, e.g., various types of matrix decompositions such as QR using Given rotations, can be agreed with the project supervisor. The project may focus more on theoretical or systems-building aspects, depending on the strength and interest of the student.
Outcome: Good understanding of the research landscape on linear algebra over relational data; Design of an algorithm that exploits the relational structure of the underlying data to lower the computational complexity of the linear algebra task; Implement the algorithm and benchmark it against existing solutions; Write-up a detailed report.
Context: Several recent MSc theses supervised by the DAST team on various matrix decompositions: QR, SVD, quadratically-regularised low-rank, give a good idea of this strand of work (pdfs available at https://fdbresearch.github.io/)
Prerequisites:
- Good programming skills in C++
- Conversant with: linear algebra; computational complexity; data structures and algorithms; databases (good grades in undergraduate courses on these topics)
- Desire to work on a hot research topic
Strand 3: Intersection Joins
Goal: Consider databases, where some colunmns can host multi-dimensional intervals instead of scalar values. Whereas on scalar values, a common notion of "agreement" is given by equality joins, for intervals one uses the intersection join that states whether the intervals overlap.
Much research has been dedicated to the evaluation of queries over databases with scalar values. In particular, research in the past decade proposed new and surprising algorithms for query evaluation that achieve worst-case optimality. They were also shown to perform much better than the standard join algorithms for a variery of settings.
This project will investigate novel algorithms for the evaluation of queries with intersection joins. The project can focus on the more theoretical or systems aspects of the problem. This is to be discussed with the supervisor.
Outcome: Good understanding of the research landscape on worst-case optimal join algorithms; Design and implementation of algorithms; Benchmarking against existing solutions; Write-up a detailed report with the findings.
Context: There are no existing results by the DAST group on this line of research. However, there is solid literature on two-way intersection joins and on worst-case optimal equality-joins.
Prerequisites:
- Good programming skills in C++
- Conversant with: computational complexity; data structures and algorithms; databases (good grades in undergraduate courses on these topics)
- Desire to work on a hot research topic
Strand 4: Incremental Maintenance of Analytical Workloads
Goal: Maintain analytics (query results, machine learning models) over relational databases under updates.
Intensive research has been dedicated to the evaluation of queries over static relational databases. One outstanding achievement in this line of work was the design of algorithms that evaluate join queries in worst-case optimal time. It was shown that these algorithms perform much better than the standard join algorithms for cyclic queries.
Incremental view maintenance considers the setting where the database is subject to frequent updates, e.g. insertions and deletions of tuples, or the data comes as a stream of tuples. It aims at providing algorithms that refresh the query result after each update as fast as possible. The naive approach that recomputes the query result after each update from scratch is too time-consuming. The literature proposes techniques like delta processing or view materialisation that perform better than the naive approach.
The DAST group built a system called F-IVM that implements a novel maintenance mechanism over fast-evolving data. Experiments showed that F-IVM outperforms state-of-the-art systems. On the theoretical side, we designed the first worst-case optimal IVM approach that maintains the number of triangles in a database in worst-case optimal time. The approach was further extended to general triangle and hierarchical queries. Partly due to our exciting recent results, incremental maintenance is currently a very hot topic investigated by several research groups within the database community.
This line of projects will investigate novel algorithms for the maintenance of different query classes like conjunctive and aggregate queries, or even more complex analytics such as machine learning models over fast evolving relational data. The project can focus more on theoretical or systems-building aspects, depending on the strength and interest of the student.
Outcome: Good understanding of the research landscape on incremental maintenance techniques; Design of a novel maintenance algorithm; Implement the algorithm and benchmark it against existing solutions; Detailed report.
Context: Recent publications and talks (pdfs and videos) available at our publications page.
- Maintaining Triangle Queries under Updates.[arxiv]
Ahmet Kara, Hung Q. Ngo, Milos Nikolic, Dan Olteanu, Haozhe Zhang. To appear in ACM Trans. Database Syst. 2020. - Trade-offs in Static and Dynamic Evaluation of Hierarchical Queries.[paper,arxiv, video]
Ahmet Kara, Milos Nikolic, Dan Olteanu, Haozhe Zhang. In PODS 2020. - Counting Triangles under Updates in Worst-Case Optimal Time.[paper, arxiv]
Ahmet Kara, Hung Ngo, Milos Nikolic, Dan Olteanu, and Haozhe Zhang. In ICDT 2019 (best paper award). - F-IVM: Learning over Fast Evolving Relational Data. [paper, arxiv, video]
Milos Nikolic, Haozhe Zhang, Ahmet Kara, Dan Olteanu. In SIGMOD 2020 (demonstration). - Incremental View Maintenance with Triple Lock Factorisation Benefits.[paper, arxiv]
Milos Nikolic and Dan Olteanu. In SIGMOD 2018.
Prerequisites:
- Good programming skills in C++
- Conversant with: computational complexity; data structures and algorithms; databases (good grades in undergraduate courses on these topics)
- Desire to work on a hot research topic
Master's Project: Implementation of an Efficient Algorithm for the Dynamic Evaluation of Hierarchical Queries [presentation]