Some of my previous work:
🧰 Data Products & Platform Development
Launched a cloud-native Data Product navigation platform (Data Mesh)
Built a product to create, manage, and utilize Data Products, enabling European clients to significantly cut costs and time spent finding relevant datasets for their use cases.
🏦 Financial Services & Risk
Forecasted stock exchange volumes to enhance service quality
Engineered scalable Spark and SQL pipelines to process 200M+ daily stock transactions. Delivered interactive dashboards that reduced reporting efforts for analysts and IR teams by 30%.
Automated underwriting for a global insurance provider
Built a reusable tool that leveraged financial, organizational, and social media data to assess customer risk profiles, reducing time-to-offer and optimizing premium pricing.
Increased Customer Lifetime Value with predictive cross-selling
Generated next-best-offer recommendations for an automotive client using customer segmentation and historical behavior data. Boosted CLV by 3% with data-driven offer timing.
🏭 Manufacturing & Supply Chain
Launched ML-powered pricing services in 24 European automotive markets
Developed and deployed models to estimate price elasticity and simulate the impact of price changes and discounts on leasing conditions in a direct B2C sales model.
Unified KPI reporting across 17 SAP systems
Built a global reporting solution for a multinational manufacturer, calculating 120+ KPIs. Applied ML models to standardize inconsistent data and ensure accuracy across heterogeneous systems.
Optimized global vaccine supply chain
Developed a just-in-time production tool matching supply, demand, and inventory across 10 plants and 4 continents. Improved coordination of hundreds of suppliers with predictive logic.
Reduced warranty costs with ML-driven EOL test optimization
Improved field quality and reduced service costs by 6% through ML-optimized end-of-line test thresholds — outperforming expert-defined values by 17%.
📰 Cost Optimization & Software automation
Cut software license costs with automated deduplication
Saved 3% in license fees by detecting unused or duplicate licenses using fuzzy matching and distance-based comparisons.
Built an AI-driven market event monitoring tool
Automated the detection and classification of business-critical news using text mining, sentiment analysis, and data fusion from multiple APIs.
🌿 Social Impact & Innovation
Reduced need for animal testing with predictive toxicity modeling
Trained and validated ML models to forecast skin irritation from cosmetic ingredients. Achieved a 22% accuracy gain over traditional animal-testing-based methods.
