I work on firmware for a robotic hand at Nibotics while studying at Leibniz Universität Hannover. Before that, I spent four years at RFmondial as a working student on digital radio broadcasting systems.
My favorite class during my B.Eng. was Real-Time Operating Systems. We worked through scenarios ranging from car airbags to satellite communication: solving equations, figuring out the RTOS behavior, and writing C with pen and paper. I’ve also always admired people who build their own electronics at home.
People bring me prototypes that need fixing or something new added—RF circuits, sensors, main boards, and motors. What looks like a software bug often ends with me checking the board with a scope or logic analyzer, or soldering a jumper wire.
As a student, I’ve participated in machine learning research projects at the L3S Research Center. For VibroSense, I worked on separating contact signals from noise in tactile recordings. In another project, I prepared tactile image data, trained models for cross-sensor representation learning and image generation, and built the training infrastructure.
Apr 2026 – Aug 2026
Semester Project, L3S Research Center, Leibniz University Hannover
Mapped DIGIT and GelSight contact images into a shared latent space using a JEPA-based ViT encoder. Trained a PixelDiT decoder for cross-sensor image generation.
Preprocessed recordings into 2 million frame samples and built the configuration-driven training harness with checkpoint/resume and distributed training.
Oct 2025 – Mar 2026
Student Project, L3S Research Center, Leibniz University Hannover
Developed a machine-learning and signal-processing pipeline for seven-microphone tactile sensor data, including the noise_separation module.
Built the training, evaluation, and inference workflow with PyTorch Lightning, Hydra, logging, checkpointing, and Slurm scripts.
SOURCE CODE