Software Engineer & Startup Founder | Berlin, Germany
Building production-grade AI/ML solutions for Industry 4.0, digital twins, additive manufacturing, predictive maintenance, and physics-informed simulation.
Creating scalable tools that drive real impact in manufacturing, quality control, and scientific discovery.
Open to freelance collaborations, framework licensing, custom development, and EU/German partnerships in digital transformation.
-
pinn-research-platform
Web-based powerhouse for Physics-Informed Neural Networks (PINNs) — 11+ modules, 200+ PDEs solved (forward/inverse, data assimilation, multiphysics). PyTorch core + FastAPI backend + interactive real-time UI. Perfect for engineering optimization & research-grade simulation. -
gnn-research-platform
Comprehensive ecosystem for Graph Neural Networks — node/graph classification, link prediction, anomaly detection, community analysis. Modular and extensible for complex network tasks. -
uninet
Unified Neural Network framework supporting classification, regression, segmentation, generative modeling, and more. Includes U-Net for semantic segmentation (2D/3D voxel/medical/image defect tasks), CNNs, Transformers, GANs, encoder-decoder, and physics-informed extensions. PyTorch-based with modular tasks and pre-built templates. -
Industry4.0-CNN-PredictiveMaintenance
Production-ready predictive maintenance pipeline using PyTorch CNNs & Siamese networks for real-time hydraulic system fault detection. FastAPI API, MLflow tracking, preprocessing → deployment.
-
AM-QADF
End-to-end Python framework for multi-modal additive manufacturing data (PBF-LB/M): voxel fusion, synchronized processing, quality assessment, anomaly detection. MongoDB/GridFS warehouse, adaptive voxels, PyVista 3D viz, full CI/CD. -
pbf-data-platform
Scalable multi-model NoSQL platform tailored for Powder Bed Fusion research — efficient storage, retrieval, and analysis of industrial AM datasets.
- AASX-Digital & aas-processor (Python + C#/.NET)
Production-grade Asset Administration Shell (AAS) platform compliant with Plattform Industrie 4.0: full AASX package processing, digital twin lifecycle management, knowledge graphs (Neo4j), vector search/RAG analytics (Qdrant), ETL pipelines, and federated learning module for privacy-preserving distributed ML (FedAvg aggregation, local training, differential privacy, secure aggregation across twins/organizations). Docker-ready deployment, modular architecture with dependency injection and testing – ideal for secure, interoperable industrial IoT digital twins and quality infrastructure systems.
- Defect-Detection Pipeline
High-performance C++23 + OpenCV pipeline for retail/manufacturing defect detection (pluggable ONNX/TensorRT backends, Docker/Conan). Extensible via Windows DLL plugin template for custom enterprise integrations.
- Core Languages — Python (expert), C++ (modern), C#/.NET
- AI/ML — PyTorch, TensorFlow, Scikit-learn, PINNs, GNNs
- Data & Streaming — PySpark, Kafka, Delta Lake, ClickHouse, ETL/ELT
- DevOps & Deployment — Docker, Kubernetes, GitHub Actions, FastAPI, MLflow, CI/CD
- Domains of Impact — Industry 4.0 · Digital Twins (AAS) · Additive Manufacturing · Predictive Maintenance · Computer Vision · Scientific Computing
Many repositories include Dockerfiles, comprehensive tests, deployment examples, and production-readiness features.
⭐ Fork, star, or explore — I'm always open to discussions on collaborations, freelance opportunities, or customizing these tools for your use case.
Let's build the future of intelligent manufacturing together. 🚀
#Industry40 #DigitalTwins #AIinManufacturing #PINNs #AdditiveManufacturing #BerlinTech