BIOGRAPHICAL SKETCH

I am a machine learning researcher with a background in deep learning, probabilistic modeling, and biomedical signal analysis. I hold B.Sc. and M.Sc. degrees in Computer Science from TU Berlin, where my research focused on uncertainty quantification, multimodal data analysis, and machine learning methods for neural and biomedical data.

My M.Sc. thesis investigated deep learning-based classification of gait modulation from chronic subthalamic nucleus local field potential (STN-LFP) recordings in Parkinson’s disease patients, with a focus on subject-independent decoding and real-time compatible machine learning methods for adaptive neurostimulation systems. This work involved the development and evaluation of CNN- and LSTM-based architectures, model distillation strategies, feature representations, and reproducible machine learning pipelines for electrophysiological data analysis.

I have conducted research at TU Berlin, Charité-Universitätsmedizin Berlin, the Physikalisch-Technische Bundesanstalt (PTB), Humboldt-Universität zu Berlin, and BIFOLD. My work has spanned uncertainty quantification and calibration for inverse problems, electrophysiological signal analysis, multimodal physiological sensing, transfer learning, and machine learning for biomedical applications. At Humboldt-Universität zu Berlin, I developed deep learning and computer vision methods for motion tracking, object detection, and behavioral analysis in laboratory mice. At PTB and Charité, I contributed to the development of benchmarking and calibration frameworks for uncertainty-aware inference in M/EEG source imaging. Within the AutoHealth project, a collaboration between BIFOLD, Charité, and the BMW Group, I develop machine learning and signal processing methods for multimodal cardiovascular monitoring using wearable and vehicle-integrated sensing systems. In parallel, I conducted research on transfer learning and data augmentation strategies for heterogeneous functional near-infrared spectroscopy (fNIRS) datasets, focusing on robust adaptation across varying sensing configurations and data regimes.

My research combines methodological machine learning with data-intensive scientific applications, with particular experience in deep learning, probabilistic modeling, transfer learning, multimodal inference, uncertainty-aware machine learning, and the development of reproducible research software.

As the founder of MathCodeLab, I also teach undergraduate courses in Computer Science and Mathematics, fostering an active learning community.

Feel free to check out my CV for more details.

RESEARCH INTERESTS

My research interests lie at the intersection of machine learning, probabilistic modeling, and scientific data analysis. I am interested in developing learning methods that generalize across heterogeneous data sources, adapt to changing environments, and provide reliable predictions under uncertainty. More broadly, my work focuses on principled machine learning approaches that emphasize robustness, data efficiency, interpretability, and statistical rigor.

Building on my previous work in uncertainty quantification, transfer learning, multimodal inference, and learning from complex biomedical datasets, my current interests include adaptive learning systems, continual and online learning, learning under distribution shifts, and multimodal data integration. Of particular interest are methods that enable efficient adaptation to new domains and tasks while maintaining robustness and generalization performance.

A central research theme is the study of learned representations and their role in generalization. This includes understanding how representations emerge, transfer across datasets and domains, and can be leveraged to improve robustness, interpretability, and adaptation. I am further interested in representation alignment, transferability of learned features, and learning frameworks that combine strong empirical performance with principled statistical foundations.

TECHNICAL SKILLS

  • Programming Languages: Python, R, Julia, MATLAB, Java, C/C++, Haskell, TypeScript
  • ML Architectures & Models: DNN, CNN, RNN (LSTM); Transformers (LLMs); Generative: GANs, AEs; Reinforcement Learning
  • ML Libraries & Frameworks: TensorFlow, Keras, PyTorch, scikit-learn, NumPy, pandas, SciPy, Numba, OpenCV, matplotlib, seaborn
  • Applied ML Domains:NLP, Computer Vision, Time Series, Signal Processing
  • Cloud & DevOps: Docker, AWS, Azure, CI/CD, RESTful APIs
  • Data Engineering: SQL, ETL Pipelines, Apache Kafka, Apache Spark
  • Software Development Tools: Git, GitHub, Bash, SSH, gdb, Unit Testing, API Design, Open-Source
  • Markup & Scripting: LaTeX, HTML, CSS, Markdown
  • Collaboration & Communication: Project Management, Scientific Writing, Teaching, Documentation

INTERESTS

Representation Learning Probabilistic Machine Learning Bayesian inference Uncertainty Quantification Transfer Learning and Domain Adaptation Online Learning Multimodal Data Integration Efficient Model Adaptation Representation Alignment Robust and Interpretable AI Cognitive-Inspired Machine Learning Generative modeling Brain Computer Interfacing (BCI) Computer Vision Deep Learning Machine Learning Natural Language Processing (NLP) Neuroimaging Reinforcement Learning Signal Processing Time Series Analysis Wearable Neurotechnology