Berlin, Berlin, Germany
I work at the intersection of physics, machine learning, and software engineering. My background in fluid dynamics and numerical methods drives my systematic approach to solving complex technical problems. After completing my PhD in two-phase flow physics, I pivoted to machine learning and software development. I've worked on large-scale sensor data processing and computer vision problems where I applied deep learning and classical algorithms to real-world problems in automotive and industrial domains.
Development and deployment of machine learning solutions in research and production environments. Technical project leadership and stakeholder coordination. Scenario-based testing platform for autonomous driving: - Designed terabyte-scale multi-sensor data processing pipelines for object detection using CNNs and Transformers - Developed digital twin reconstruction of road scenes using adapted SLAM algorithms - Built attention-based models to analyze stochastic dependencies in binary encoded sensor data - Development in Python and C#, production deployment on AWS infrastructure Anomaly detection in industrial sensor data: - Analysis of 100+ dimensional time-series for long-term battery tests (terabytes per cycle) - Implementation and evaluation of fast vector-quantization-based semi-supervised learning methods - Development in C# and Python for on-prem deployment and automated monitoring Quality Assurance in Medical Imaging: - Defect classification of histopathological images using convolutional neural nets - Optimization with sampling techniques for high-speed analysis of gigapixel images LLM-based climate risk analysis: - Climate risk analysis for CSRD reporting using LLMs - Agent implementations for multilevel analysis and climate data retrieval - Deployment and hosting using Azure Services Team management: Led a machine learning team of 5 engineers within a 30+ member automotive project Architected computer vision pipelines processing multi-sensor data for autonomous vehicle testing Coordinated with product owners and stakeholders on technical roadmap and delivery
Data analytics, and computational fluid dynamics simulations. Numerical modeling of fluidic nozzle systems. Smart-Nozzle Hub: - Smart nozzle sensory data management system - Front- and backend development in Python/PyQt - SQLite database integration Automated droplet analysis of atomizing liquid sprays: - CNN-based droplet detection and segmentation tool - Front- and backend development in Python/PyQt, SQLite integration
Computational methods for the linear stability of two-phase shear flows. Thesis: On the development and application of linear stability methods for two-phase shear flows. - Developed a framework for global linear stability analysis of 3D two-phase flows. - Implemented high-accuracy linear solver in C using VoF and Level-Set methods - Simulation and analysis of large-scale flows on HPC clusters.
Tasks: - Numerical simulation of diffusion flames - Numerical simulation of industrial scale LPG leakage
Tasks: -Investigating the potential of the C++ FEM library deal.II -Development and Optimization of a FEM code for modeling two-phase flow in porous media with adaptive mesh refinement