At Fruugle, I cleaned, categorized, and preprocessed 50K+ retail product records for consistent downstream machine learning workflows. I also labeled ground-truth data and extracted features using defined guidelines.
I ran structured QA testing on a mobile application and documented defects and improvement points in actionable feedback reports. At Bitycle, I developed and fine-tuned Hugging Face transformer models for cryptocurrency news sentiment analysis and named entity recognition; the work supported a customer-facing news and social-media signals feature.
At Sajayanegar, I benchmarked international regulations and practices, including GDPR and the EU AI Act, to support development of a national data governance framework. I prepared materials translating technical and regulatory findings for policymakers and collaborated with them on data sovereignty initiatives.
In projects, I designed interactive Power BI dashboards with DAX-based analytical models and built an interpretable XGBoost insurance-cost prediction pipeline, using SHAP and causal residualization to reduce age-related bias. I’m pursuing an M.Sc. in Data Science for Economics at the University of Milan, where my thesis examines bias in LLM-based multi-agent social media simulations.

