Franziska Ahrens

Built end-to-end AI/ML solutions for international humanitarian org and enterprise clients on GCP. Developed production ML pipelines with Airflow orchestration, data lineage tracking, and anomaly detection systems. Architected centralized GCP data serving layers for recommendation/personalization use cases, deploying Streamlit apps with BigQuery integration and data quality checks. Established engineering standards and promoted data-as-a-product mindset across teams.

Led facial similarity project within large-scale skincare study using computer vision models to generate face embeddings and identify lookalike participants. Addressed data quality challenges from image resolution/lighting variability through ML-based preprocessing. Compared skin metrics across matched participant groups to improve skin-personality clustering. Technologies: Python, Polars, Azure, SQL, computer vision models.

Contributed to feed relevance ranking: integrated real-time features, retrained model, evaluated with SHAP—deployed to production with engagement gains. Ran A/B experiments optimizing content diversity and new-user onboarding. Prototyped intelligent job search using GPT-4 and BERT embeddings, inspiring XING's flagship search feature. Facilitated cross-team integration across 15-developer organization. Technologies: Python, Spark, SQL, MLflow, Docker, OpenAI API, BERT.

Integrated two data teams into a cohesive 15-developer unit during critical transition. Led ORSC-facilitated workshops to establish new collaboration processes and foster trust. Represented the team in strategic management meetings and facilitated retrospectives. Designed agile methodology workshops across teams and helped build an internal agile community of practice.

Nachbereitung der Anlageninventur