WORK

Project library

A focused collection of data analysis, experiment design, dashboarding, and product analytics work.

EEG Ad Recall Prediction Thesis Cover; Polimi Logo; Brainwave Patterns; TV Ad Icons; Data Science Graphics

My website: A portfolio built with agentic AI tools

Built this portfolio as a website with agentic AI tools, using Codex, Claude, ChatGPT, Git, and Vercel instead of a no-code website builder.

CodexClaudeChatGPTVercel
Portfolio hero image for the Superstore Sales Performance Dashboard, showing a Power BI executive overview dashboard with sales, profit, regional performance, and profitability matrix charts.

Superstore Sales Performance Dashboard

Built an executive Power BI dashboard on the Superstore dataset for quarterly sales, regional performance, categories, and customer segments.

Power BIDashboarding
Experiment analysis graphic showing bounce-rate improvement, guardrail checks, and device-level treatment effects for a landing page A/B test.

Maven Landing Page A/B Test Analysis

Analyzed a landing page experiment with traffic filtering, bounce-rate testing, guardrails, and treatment-effect cuts.

PythonA/B testing
EEG Ad Recall Prediction Thesis Cover; Polimi Logo; Brainwave Patterns; TV Ad Icons; Data Science Graphics

Predicting Advertising Recall from Brain Signals

My M.Sc. thesis at Politecnico di Milano, I built a machine learning pipeline that predicts whether a TV advertisement will be remembered - before it ever airs - using EEG brain signals recorded from viewers.

PythonMachine learning
Technical cover showing a modular Titanic ML pipeline with config, validation, feature engineering, model, and prediction blocks.

Titanic ML Production Pipeline

Turned the Titanic notebook workflow into a modular Python ML system — configuration-driven, validated, with custom transformers, persisted artifacts, and a clean prediction interface.

PythonMachine learning
Technical cover showing a Notebook ML Prototyping path with EDA, Feature Engineering, Modeling and comparisons.

Titanic Survival Prediction Machine Learning Research

Used the Kaggle Titanic dataset as a controlled machine-learning exploration sandbox. Discovered insights through EDA, engineered defensible features, benchmarked 8 model families, and selected Gradient Boosting for its stability-performance tradeoff.

PythonScikit-learnXGBoostCatBoostPandas
Power BI dashboard showing sales, brand health, and digital visibility KPIs for a beverage client.

Marketing Performance Analytics for a Beverage Client

Designed a KPI framework and Power BI dashboard for a global beverage client, integrating retail, search, and social data into a single performance view for marketing leadership.

PythonPower BI

COLLABORATE

Open to building something useful together?

If you want a data or product partner for a thoughtful project, I'm open to the conversation.

Start a conversation