2017 - 2025: Quality Management

Einsatzort: Berlin

Kunde: Pharmaceutical Industry

Aufgaben:

• Data-driven process optimization in regulated environments
• Development and improvement of structured workfl ows and documentation systems
• Strong focus on validation, reproducibility, and reliability
• Collaboration with interdisciplinary scientifi c and regulatory teams

2005 - 2024: Road Accident Forecasting (Spatio-Temporal Modeling)

Aufgaben:

• Processed large, noisy and zero-infl ated real-world datasets
• Developed and evaluated predictive models (Logistic Regression, GNN)
• Built feature pipelines including spatial aggregation (H3) and temporal features
• Systematically evaluated model performance across multiple resolutions

various Projects

Aufgaben:

Agentic PDF Parsing (AI System Architecture

• Designed and implemented a modular AI system combining LLM-based extraction with deterministic logic
• Built structured pipelines for multi-document analysis and cross-document reasoning
• Implemented validation, caching, and retry mechanisms to ensure reproducibility and robustness

Escalation Detection System (LLM vs ML; missing final wrap-up)

• Developed an end-to-end pipeline for structured text classifi cation (rule-based, ML, LLM)
• Built evaluation framework for comparing model performance and decision consistency
• Integrated LLM outputs into reproducible decision workfl ows

X-Ray Classifi cation (Deep Learning)

• Built CNN-based classifi er (ResNet50, transfer learning)
• Applied explainability methods (Grad-CAM, SHAP) for model interpretation

Multimodal Recommender System

• Developed end-to-end pipeline combining text and image features
• Implemented ranking logic and evaluation workfl ows

Financial Data Analysis & Clustering

• Collected and processed large-scale fi nancial datasets via APIs
• Applied PCA and clustering to identify patterns across assets