At Hahn-Schickard, I built an end-to-end machine learning pipeline that converted bio-acoustic signals into CWT spectrograms. I compressed autoencoders to under 150 KB for deployment on microcontrollers.
I developed sensor-specific models and tuned the low-cost MEMS model to nearly match the model trained on high-quality condenser microphone data, eliminating the need for expensive measurement hardware. I also engineered an evaluation metric tailored to real-world monitoring that retained most target signals.
I completed an M.Sc. in Biomedical Engineering at Hochschule Anhalt, with a thesis on detecting plant ultrasonic acoustic emissions using machine learning. My focus areas included machine learning, biomedical signal processing, medical imaging, modeling and simulation, and computer-assisted medicine.
