Summary
Stefan Taubert is a research data scientist and software engineer with a PhD in neural speech synthesis and nine years of experience building and evaluating ML systems across academic and applied settings. He develops, trains, and optimizes deep learning models (PyTorch/TensorFlow), implements multimodal pipelines (text, audio, image), and deploys performant solutions on HPC and GCP, while maintaining 20+ Python packages and open-source releases on PyPI, Zenodo, and HuggingFace. His work spans TTS evaluation (subjective MOS and objective metrics), prosody-aware encoder–decoder extensions, transfer learning, and efficient edge inference—evidenced by a 10x real-time BirdNet acceleration on a Raspberry Pi 3A+. Awarded for both research and application (CLEF placement, IEEE paper prize, Deutschlandstipendium), he also brings hands-on leadership from supervising research assistants and teaching university courses to mentoring a master’s thesis. Colleagues benefit from his interdisciplinary collaborations in linguistics and psychology and his knack for reducing manual effort via NLP-driven pipeline optimizations.
9 years of coding experience
12 years of employment as a software developer
Abitur, Abitur at BSZ für Technik II
high-school diploma, high-school diploma at Untere Luisenschule Chemnitz
PhD (Dr. rer. nat.) Computer Science, PhD (Dr. rer. nat.) Computer Science at Technische Universität Chemnitz
English, German, French