writing thesis in progress...
About
I am a data science research assistant with a background and passion for software engineering. My interests include, but are not limited to:
-
Software Development
-
Sports (Handball, Cycling, Sailing & Mountaineering)
-
Food & Beverages
-
Photography
Projects
The goal of the LINA consortium is to create a basis for the largest European (real and virtual) infrastructure for research, development and safe testing of autonomous systems / UAS (Unmanned Aircraft System) for commercial products in the canton of Zurich.
Childhood injuries, a significant concern lacking Swiss data, prompted a study assessing machine learning (ML) to extract injury data from electronic records (2018–2022, N=30,884) at Zurich’s pediatric emergency department. Steps included evaluating text utility, developing ML models, and testing performance. Results confirmed data feasibility, with adaptations to coding systems; preliminary findings showed 36% of injuries in ages 0–4, 58% male patients, and 73% mild cases. ML can enhance injury monitoring if digital records and detailed inputs are available.
Key Points:
-
ML effectively extracts injury data from pediatric records, addressing Switzerland’s data gap.
-
Requirements: Digital access, tailored coding systems, and thorough clinical documentation.
-
Insights: Young children and males are most affected, with severe injuries being rare (<1%).
Side Projects
Speech Translation models designed to convert spoken Swiss-German to written German have been in existence for some time. While these models generally perform well, their performance in various scenarios remains poorly understood. In this thesis, we explore the influence of audio length on the performance of Swiss-German speech translation models and identify the necessary factors for achieving better performance on longer audio segments. To achieve this, we examined four speech translation models from different institutions. A model from the Zurich University of Applied Sciences (ZHAW), one from the University of Applied Sciences Northwestern Switzerland (FHNW), a model from Microsoft, as well as a model from OpenAI called Whisper. We conducted eight different experiments using a Swiss-German corpus collected by the ZHAW and FHNW. In the experiments, the audio length was augmented in various ways. From there, we found that while the ZHAW, FHNW and Microsoft models showed a tendency to perform worse on longer duration, extending the duration by adding silence did not influence on the performance. Changing the playback speed has a negative influence on the ZHAW, Microsoft and Whisper models, both when speeding segments up or slowing them down. The FHNW model exhibited extraordinary robustness to changes in playback speed, as the results when accelerated by a factor of 1.25 were nearly identical to the results when the playback speed was not altered. The biggest influence on performance was when adding more than one sentence to a segment. Without a segmentation of the input audio the ZHAW, FHNW and Microsoft models performed badly, indicating that segmentation should be introduced as soon as more than one sentence appears in an audio recording. Training a model specifically on multi-sentence segments showed promising results, on single sentence segments and multi-sentence segments as well as in scenarios where sentences are split while segmenting the audio recordings. Comparing a sentence-based segmentation, which is considered ideal for models trained on single sentence segments, to a fixed-window segmentation with an overlap showed an almost identical result. Examining the models on a real-life recording showed that the ZHAW (lowercase) and ZHAW (multisentence) models perform considerably worse than the FHNW, Microsoft and Whisper models. Indicating that more investigation is required to fully understand what makes a speech translation model work well in real-life scenarios.
Let your friends know when and where you are about to drink your favorite beverage and invite them to join you. In a busy world where no one has time to meet we help you connect with your friends again.
MDTeX allows to write math formulas using TeX inside Markdown and preview them using the Visual Studio Code Markdown previewer.
A calculator for logic expressions and a tool to turn truth tables into logic expressions.
A realtime tracking dashboard to visualize geo locations using firebase and GeoJSON.
A tracking dashboard to visualize geo locations of trucks in Nigeria. This was part of the LafargeHolcim challenge at HackZurich 2020.
A platform where farmers can show the products their farm produces and sells. Users can order products which then will be reserved for them to be picked up.
The goal was to let people know which farm is selling which products and to encourage them to go to the farms to get these products.
A tool to evaluate the mood of faces recognised by one or more cameras continuously in real time.
Work Experience
I am a member of the Explainable AI Group (XAI) at the Centre for Artificial Intelligence within the ZHAW School of Engineering. In this role, I spearhead the implementation of machine learning initiatives, focusing on delivering robust and interpretable AI solutions. Additionally, I contribute to software engineering efforts, ensuring scalable and maintainable system architectures to support research and development objectives.
As an independent consultant, I helped end-to-end Proof of Concept (PoC) development for cross-platform mobile applications (React Native) and full-stack web applications (React/Node.js). Additionally, I provide consulting services to clients, advising on software development frameworks and data science-driven solutions.
I was hired by Buhler Group AG through Bluematic AG. There I am responsible for developing process
automations within the Azure DevOps platform. I’m also working in a team to unify frontend components in an
angular library.
My responsibilities included developing API’s using PHP (Symfony 4) and MySQL. During this time, I was
mentoring two interns. The biggest challenge at that time was building an App in React Native and a Front-end
for a large e-commerce site. I was also responsible for various DevOps tasks such as building CI and CD
automations.
The main responsible was front-end development. One of my biggest achievements there was to implement a
new development stack (React, Redux, SASS, Webpack) and educating my coworkers on it.
Responsible for the development of a tool that allowed our customers to migrate data from different ERP
interfaces into the web-based platform, the company was developing at that time, without the need for a
redeployment. This allowed us to save a lot of hours spent by developers building new interfaces and
deploying the application on customer environments.
During my apprenticeship I was part of a SCRUM team with members in Poland and Switzerland. Together we
worked on a web-based platform for after sales services that is used by industrial customers. My
responsibilities included the mobile optimization of the platform, implementing new workflows and building
small data processing tools.
Volunteering
Every year I spent one week as a coach in the kids camp organised by the Evangelische Reformierte Kirchgemiende.