Data Scientist · AI / ML Engineer · Computational Physicist
Lead AI developer at DealCircle GmbH · Open-source author (TelescopeML) · Explainable AI (XAI) specialist
I am a senior data scientist and machine learning engineer with a Ph.D. in computational physics and more than 10 years of experience turning data into reliable, production-grade systems. My work spans the full lifecycle of AI/ML projects, from research and prototyping to deployment, in close collaboration with cross-functional teams.
- 🔭 Lead developer of recommendation systems and AI applications at DealCircle GmbH.
- 🧠 Specialist in Explainable AI (XAI), improving the transparency and trustworthiness of black-box models.
- 📦 Author and maintainer of the open-source Python package TelescopeML, including its XAI module.
- 🚀 Builder of end-to-end AI / ML / Deep Learning projects across computer vision, generative models, and time-series.
Explainable AI (XAI) · Deep Learning · Computer Vision · Generative Models · Recommendation Systems · Time-Series Analysis · MLOps & Deployment
Languages
ML & Deep Learning
Data & Visualization
Backend & Deployment
| Project | Techniques | Data Types | Poster |
|---|---|---|---|
| Search Semantically — LLMs & Semantic Search Handbook A 14-chapter open-source course with runnable notebooks, a free e-book and a docs site: from keyword search to embeddings, dense retrieval, re-ranking, evaluation and RAG. |
Embeddings, Dense Retrieval, BM25, Vector Databases & ANN, Re-ranking, RAG, Search Evaluation, MkDocs, CI | Text, Vectors |  |
| Agentic System Design Tutorial A step-by-step guide to architecting a production-grade agentic AI system — LLM gateway, orchestrator, sub-agents, tool registry, memory, guardrails and observability. |
Agentic AI, LLM Orchestration, Tool Calling, Memory & RAG, Guardrails, Observability, System Design | Text, Architecture Diagrams |  |
| The 12-Factor App — Explained for Beginners An interactive, single-file tutorial on the 12-Factor methodology with plain-English analogies, Python do/don't examples and hand-drawn diagrams. |
Cloud-Native Design, 12-Factor Methodology, Docker, Kubernetes, CI/CD, Software Architecture | — |  |
| Pub/Sub Best Practices Tutorial A visual beginner's guide to Google Cloud Pub/Sub best practices — batching, ordering, pull vs. push, dead-letter topics, IAM and testing — with console mockups and Python snippets. |
Messaging Systems, Google Cloud Pub/Sub, Event-Driven Architecture, Reliability Patterns, IAM | Streaming Messages, Events |  |
| SOLID Design Principles Tutorial Learn SOLID design principles through interactive Jupyter notebook tutorials. |
SOLID principles, design patterns, OOP, software architecture, code quality | — |  |
| TelescopeML — Open-Source Python Package Deep CNNs and ML models for analyzing stellar and exoplanetary telescope spectra. |
Deep CNN, Machine Learning, XAI, Bayesian optimization, Feature Engineering | Timeseries, Tabular |  |
| Malaria Cell Classifier Deep CNNs and ML models for anomaly detection in microscopic malaria cells. |
Deep CNN, Data Augmentation, Feature Engineering, Image Processing, Optimization | Image |  |
| BloodPy — Automated Blood Cell Classifier Multi-classification of peripheral blood cells using deep CNNs and ML models. |
Deep CNN, Data Augmentation, Transfer Learning, U-Net, Image Processing, Statistical Analysis, OpenCV, Fine-tuning | Image, Metadata |  |
| Dataset: Segmented Peripheral Blood Cells (OpenCV) A dataset of segmented white blood cell images using advanced image processing. |
GrabCut, Morphological Operations, OpenCV | Image, Binary Masks, Dataset |  |
| LIME for Macroscopic Medical Images A surrogate model (LIME) for enhancing transparency of medical diagnostics. |
Deep CNN, LIME, XAI, Computer Vision, Optimization | Image |  |
| CAM for Macroscopic Medical Images Class Activation Mapping (CAM) for anomaly localization interpretability. |
Deep CNN, CAM, XAI, Computer Vision, Data Analysis, Optimization | Image |  |
| Automated Nucleus Detector A semantic segmentation solution for automating nucleus detection in biomedical images. |
U-Net, Keras-Tuner, Semantic Segmentation | Image |  |
| FastAPI Questionnaire API A FastAPI application to manage and retrieve questionnaire data with authentication and custom error handling. |
FastAPI, Authentication, Data Management, Error Handling, Shell | API, CSV |  |
| Neural Compression Advanced autoencoder architecture for efficient lossless data compression. |
Autoencoder, GenAI, SSIM, PSNR | Image |  |
| LIME for ECG Classification A surrogate model (LIME) for enhancing transparency of time-series classification. |
CNN, LIME, Time Series Analysis | Timeseries |  |
| Beta-Variational Autoencoders Generative learning (GenAI) with Beta-Variational Autoencoders. |
Beta-VAEs, Latent Space Analysis | Image |  |
| Movie Recommendation Systems Collaborative filtering, content-based filtering, and Bayesian average techniques. |
Collaborative Filtering, Content-Based Filtering, Bayesian Average | Metadata, User Ratings, CSV |  |
| Falcon 9 Rocket Predictor A data-driven project to predict the success of Falcon 9 rocket landings. |
Data Wrangling, Feature Engineering, Web-Scraping, JSON Processing, SQL, Hadoop, Folium, Decision Trees | Unstructured, Tabular, Database |  |
| Variational Autoencoders A collection of VAE architectures built with the Keras deep learning framework. |
Variational Autoencoders, Exploratory Data Analysis | Image |  |
| Netflix Content Analysis Exploratory analysis of Netflix's catalog to uncover trends in content distribution and quality. |
Exploratory Data Analysis, Data Visualization, Statistical Analysis | Metadata |  |
| Air Passenger Time-Series Analysis Exploratory analysis of the Air Passengers dataset, uncovering trends and patterns over time. |
Time-Series Analysis, Log Transformation, Moving Averages, Seasonal Decomposition | Timeseries |  |
Open to collaboration on AI, machine learning, and explainable-AI projects. Feel free to reach out.