At BIP Group, I developed a multi-agent system that checks project documentation against Responsible AI compliance controls, returning a verdict and supporting evidence for each requirement. I built its RAG pipeline with an LLM-as-judge and reviewer step, delivered as a FastAPI service on OpenShift.
At BIP Group, I also designed an end-to-end RAG system for semantic search and classification of PubMed scientific abstracts, supporting clinical research in the cleft lip and palate orthodontic domain. For data quality monitoring, I deployed machine learning pipelines for anomaly detection and time series forecasting.
At Technology Reply, I engineered PySpark ETL pipelines for banking data and designed data models for analytics-oriented warehouses. Earlier, as a Junior Data Scientist intern at Crisma SRL, I built credit card fraud detection models using PySpark ML.
