Bio


- 2011–2015 · B.Sc. Applied Computer Science · TU Dortmund
- 2016–2018 · Involvr · Co-founder & VR developer
- 2018–2020 · Master's & freelance · VR and 360-degree web applications
- 2021–2023 · Opta Data Finance · Machine learning engineer
- 2023–2025 · Machine Learning Reply · Senior consultant
- 2026–present · Independent · Own products & selective freelance


Volvo Germany
6 monthsStep inside the Volvo in VR
At Involvr, our team created immersive VR experiences for the Volvo XC40 and XC60, including fully modeled interiors and exteriors.
- Co-founder, developer, and client delivery

Opta Data Finance GmbH
3 yearsFrom notebook to production scale
I helped build the machine learning platform behind large-scale document processing for a healthcare billing provider.
- OCR, object detection, and classification at production scale

Airbus
1 yearPredictive maintenance from flight data
For Airbus Defence and Space, I worked with aircraft sensor data to support predictive maintenance for the A400M.
- Terabytes of time-series data with PySpark

FRoSTA
1 yearQuality control on the production line
At FRoSTA, I designed an edge computer vision system for automated quality control in frozen vegetable production.
- Industrial camera and hardware integration

Red Bull
3 monthsAI face swap at stadium scale
For Red Bull, I built a scalable head-swap service that integrated generative AI into an existing fan application.
- Stable Diffusion service with content moderation

Eventim
1 yearOne product, twenty markets
At Eventim, I managed a web-based analytics product used across more than 20 international markets.
- Product strategy across more than 20 international markets


AI and data
Models, pipelines, and inference
- TensorFlow for OCR, object detection, classification, and clustering
- Stable Diffusion, content moderation, and GenAI video generation
- PySpark pipelines for terabyte-scale time-series and forecasting
- Edge computer vision on Nvidia Jetson for industrial quality control
- Vercel AI SDK agents with tool calling, chats, and endpoint protection
Systems
Production engineering
- Next.js, TypeScript, React, shadcn/ui, and Tailwind
- Supabase and Convex for auth, data, and realtime backends
- Python and Java (Spring Boot) microservices with REST, gRPC, and JMS
- Kubernetes, Docker, AWS CDK, ECS, and on-prem GPU deployments
- C++/OpenGL engine compiled to WebAssembly for immersive VR experiences
Operations
Keeping it alive
- Sentry for error tracking and production observability
- AI-assisted CI/CD pipelines and agent-orchestrated development workflows
- Prometheus, Grafana, and Loki for monitoring, dashboards, and log analysis
- Paddle for subscription billing and payment integration
- Skill management, context engineering, and token-efficient agent design
Growth
Shipping and learning in public
- Designing and managing Google Ads campaigns for product validation
- Social media marketing and competitive positioning content
- Building in public - sharing progress, learnings, and product decisions openly
- SEO comparison pages, launch offers, and funnel iteration for micro-SaaS
- End-to-end product ownership from prototype to paying customers
Bachelor thesis · 2016
Mining of geo-tagged Word-Clouds from bibliographic data in computer science
Built a pipeline to cluster, summarize, and visualize bibliographic computer science publications as geo-tagged word clouds - combining NLP preprocessing, TF-IDF weighting, and spatial mapping of research topics.
- NLP preprocessing, clustering, TF-IDF, and topic modeling
- Geo-tagging and visualization of bibliographic research data
- Grade: 1.3 · TU Dortmund, Applied Computer Science
Master thesis · 2020
Fast Latent Dirichlet Allocation with Background Topics
Designed and benchmarked a runtime-optimized sampling algorithm for Latent Dirichlet Allocation with background topics - improving inference speed while preserving model quality on large text corpora.
- Probabilistic topic models with background-topic handling
- Java implementation with Python analysis (SciPy, scikit-learn, gensim)
- Grade: 1.0 · TU Dortmund, Applied Computer Science