Research
Research Projects
Reliable DNN-Based Software using Supervisors (PhD Thesis)
As the use of Deep Neural Networks (DNN) becomes increasingly prevalent in software systems, it’s important to note that these systems may not always produce the desired results, which can lead to failures in systems that rely on DNNs. This is particularly concerning in safety-critical systems such as self-driving cars and medical applications.
Here, my research centered on the design, development, and analysis of tools and methodologies to detect and handle uncertainties and errors that may arise from the use of DNNs in software. By providing software engineers with our results and tools, I aimed to help them create more robust and reliable systems that utilize DNNs.
This work (which formed my PhD thesis) was part of the ERC-funded Precrime Project, led by Professor Paolo Tonella at the Università della Svizzera italiana.
A Large-Scale Graph of Software Dependencies and their Licenses
This 6-month project (scheduled to start in May 2023) will provide a comprehensive and quantitative overview of the license compatibilities between commonly used open-source software packages, which directly imply legal risks to vendors of proprietary software. It furthermore allows to reveal the typically unin tentional yet illegal use of transparently included upstream dependencies under insufficient licenses. To achieve this goal, due to the number and high connectivity of different open source software releases, this project poses major technical and ops difficulties.
The project, for which I act as Principal Investigator, is funded by a Hasler Research Grant I was awarded 6 months before concluding my PhD.
Computational Market Design and Social Choice
I have a strong interest in the intersection of economics and computer science: I am convinded that abstract concepts such as Market Design, Algorithmic Game Theory, and Social Choice become highly relevant in modern, algorithm-driven systems, where rational agents act as proxies aiming to maximize various actors utilities - and where systems have to be designed to take these agents into account.
To that extent, I investigated the manipulation strategies in one-sided matching markets, such as school-choice and allocation problems, and explored the simulation of computationally complex combinatorial spectrum auctions. Parts of this work was conducted as part of my BSc and MSc studies at the University of Zurich and the University of New South Wales Parts. The results thereof were published at IJCAI and AAMAS.
Publications
My research is published at a range of conferences and journals, amongst which are top-tier venues like IJCAI, AAMAS, EMSE and ICSE. A list of publications I contributed to can be found on Google Scholar.
Honours and Achievements
- Received the Best Paper Award at ICST 2021 for the paper “Uncertainty Wizard” paper in the testing tool track
- Achieved the highest grade among all MSc in Informatics graduates at the University of Zurich during the period of Feb’16 to Oct’17
- Invited to present my MSc thesis at INFORMS 2016 conference in Nashville, USA
- Received top grades for my BSc and MSc thesis
Grants
- 2022: Research grant for the proposal “Detecting Open-Source License Conflicts in Software Dependencies” by Hasler Foundation (CHF ~50’000)
- 2015: Grant for a study and research exchange semester at UNSW / Data61 in Sydney (CHF 4’500)
- 2012: Grant for a exchange semester at Radboud University Nijmegen, NED (CHF ~1’500)
Advisors
I am extremely grateful for the help and support of great advisors:
- Phd Thesis (Università della Svizzera italiana): Prof. Paolo Tonella, Dr. Andrea Stocco
- Msc Thesis (University of Zurich): Prof. Sven Seuken, Prof. Ben Lubin, Dmitry Moor
- Msc Research Project (University of New South Wales / Data61): Dr. Haris Aziz, Dr. Nick Mattei, Prof. Toby Walsh
- BSc Thesis (University of Zurich): Timo Mennle, Prof. Sven Seuken