I am a Master student in Visual Computing at the Saarland University currently doing my Thesis at DFKI Saarbrücken on tracking of reaction-fronts in microscale combustion experiments. I have 5+ years of experience mainly working with Python on projects in industry and academia. I am interested in classical and deep learning-based approaches to computer vision, image processing, and computer graphics.
Oct. 2022 - now
Current grade: 1.7
Sept. 2021 - Sept. 2022
Final grade: 1.3
Okt. 2017 - Aug. 2021
Final grade: 2.2
Apr. 2024 - Now
Tracking of Self-Propagating Reaction Fronts in Reactive Multilayer Foils based on high-speed Microscopy Recordings.
Mai. 2024 - Now
Working with a team of 12 people on augmenting the rendering framework "Lightwave" of the Computer Graphics chair.
Oct. 2023 - Feb. 2024
Design and Implement Neural Networks in PyTorch to predict agricultural yields from hyperspectral satellite images.
Mai. 2022 - Sep. 2023
Design and Implement Neural Networks in PyTorch for Object detection and segmentation.
March. 2020 - Sep. 2021
Processing and analysis of medical data in Python and classification with Random Forest.
Development of a web application for using the model in a clinical setting.
Python, C/C++
PyTorch, OpenCV, NumPy, Pandas, Scikit-Learn, Scikit-Image
Git, CMake, VS Code, PyCharm, CSS, HTML
I like learning languages. Apart from my mother language German and English, I speak fluent French, Spanish and Lebanese Arabic.
I'm into strength training. I've built my own home gym with a power rack, adjustable bench and various bar- and dumbbells. I'm currently training 4 times a week.
I'm interested in the philosophy and history of liberalism. I especially like the works of Friedrich Hayek.
Implemented a ray tracer in C++ with additional postprocessing and other features. Created a scene in Blender and rendered it with our own ray tracer.
We created a data-generation pipeline that leverages Facebook AI's Segment Anything model to cut out semantically meaningful parts of images and inpaint them with Stable Diffusion. We then trained a classifier on the dataset to decide if an image contained inpainted regions.
Implemented a fully connected neural network in Python from scratch with backpropagation and gradient descent. The networks depth and width can be adjusted by the user.