Deployed Solutions
TruthLens: Forensic Fake News Detection
A multi-layered verification platform combining BERT Deep Learning, heuristic analysis, and BERTopic modeling. Unlike standard classifiers, TruthLens focuses on forensic text analysis—cleaning "data artifacts" (UI leaks, headers) so models learn semantic untruths rather than metadata shortcuts.
- Robust Cleaning: Removes temporal markers & leaks.
- Live Dashboard: Streamlit app with auto-model download from Hugging Face.
Inflation Forecasting in Crisis
Evaluated 9 forecasting models (MLP, Ensemble, Bayesian DLMs) during economic shocks. Implemented Sequential Monte Carlo for regime-switching dynamics.
Read Analysis →Aircraft Delay NLP
End-to-end pipeline converting free-text pilot logs into structured data. Fine-tuned BERT models achieved 1.00 F1 on delay type classification.
See App →Automated Inventory
Warehouse automation using an ensemble of EfficientNet-B3, ResNet-50, and ConvNeXt to count pallets.
Case Study →NobelBet Platform
Full-stack betting platform for Nobel Prizes. Handles real-time user bets and live odds.
Price Intelligence
Scraper for Back Market. Handles JS-rendering to capture historical pricing trends.
View Code →Medical AI Diagnosis
XGBoost classifier pipeline with SMOTE, RFE feature selection, and GridSearch optimization.
View Code →Climate Forecasting
Bayesian Dynamic Linear Models & Hidden Markov Models (Gibbs Sampling) for temp trends.
Satellite Classification
Terrain classification using Random Forest & XGBoost on satellite imagery.
View Code →Professional Experience
Finance Analyst Intern (Data & Automation)
Amazon Air EU • Luxembourg
- Engineered end-to-end ETL pipelines (Python/SQL) for daily reporting, saving 10+ hours/week.
- Developed scripts for 50+ KPI calculations and optimized route efficiency.
- Built Amazon QuickSight dashboards with 20+ visualizations, improving reporting clarity by 10%.
Business & Technology Specialist
De Giorgi Cosimo & Figli • Italy
- Developed a custom Warehouse Management System to optimize slow-rotation goods storage, reducing picking time by 10%.
- Conducted market research and supported strategic planning.
Education
MSc Economic & Social Sciences
Bocconi University • Milan
- Relevant coursework: Algorithmic Decision Making, Computer Science (PhD), Empirical Industrial Organization, Time Series Analysis, Advanced Math & Statistics, Applied Machine Learning, Big Data.
Exchange Program
Georgia Institute of Technology • Atlanta
- Selected on merit. Focus: Technology & Operations, Corporate Finance.
BSc Economics & Management
Bocconi University • Milan
- Grade: 110/110. Thesis on Monetary Policy & COVID-19.