Top Tips is a lightweight guessing game for handball leagues. You can choose from more than 130 leagues in 25 countries as well as international leagues.
You can create groups for different leagues and seasons and play with your friends.
The app was built with Vue and Ionic. You can find the app on Google Play or on the web.
This project encompasses autoencoders, more specifically denoising autoencoders. These autoencoders were used for a paper during my Bachelor studies and were trained to reconstruct (noisy) MNIST images. Tensorflow (i.e. keras) was the main library used for autoencoder construction and training.
ktlint is a simple linter for Kotlin. There are many built-in linting reporters (with different output formats) and several custom reporters for ktlint. We used GitLab and Kotlin in a university project, however no correctly formatted reporter existed, so I decided to write this one. This correctly formatted output can be used in GitLab CI to automatically highlight lint errors with information like file or line number as seen above.
As an exam substitute for a Text Analytics course we explored semi-automatic analysis of political parties, in this case major German political parties. We used the party manifestos and extracted topics using LDA and HDP, as well as BERT. All paragraphs of the manifestos were then categorized and each category summarized.
As an easy way to interact with the results of the aforementioned text analytics project, we created a web interface. In this interface, each category displays anonymized summaries for each analyzed party. The user can rate these summaries and in the end receives a ranking of what party the user agrees most with.
I created a few Discord bots notification system or easy role assignment on servers.
I coded a script and webpage to assign workspaces based on preferences and availability. You can assign workspaces as `important` to always fill these spaces.
As part of my Bachelor's Thesis, I worked on and extended a project to automatically extract and grade statistics reported in scientific publications.
During my studies I worked on several tasks concering Deep Learning using, for example,embedding models, CNNs, RNNs, and DCGANs. Some tasks include reinforcement learning, image segmentation, natural language inference, image synthesis, etc.
I built a small application to learn capital cities of flags of countries. You can also learn the Greek alphabet. The app is work-in-progress, but I plan to publish it soon.
As a part of my research assistant job, I was involved in many projects concerning LLMs with the focus of question answering. I looked at the capabilities of LLMs when answering moral dilemmas, factual vs. multiple choice questions, providing an answer even when that answer is incorrect, and the false confidence reported by LLMs. Most of these projects are not public, but I'd be happy to have a chat with you about these topics. Just reach out down below.