Portfolio
A selection of data analysis, machine learning, and visualization projects — ranging from production ML systems to exploratory data storytelling.
A machine learning pipeline predicting subscriber churn with 89.4% accuracy, using gradient boosting on behavioral and demographic features. Deployed as a FastAPI microservice consumed by the CRM team.
An interactive web visualization exploring 140 years of global temperature anomaly data, built with D3.js and Python. Features animated time-series, choropleth maps, and cross-regional comparisons.
A fine-tuned BERT model for aspect-based sentiment analysis on product reviews, processing 50K+ reviews daily through a streaming pipeline built on Apache Kafka and deployed on Kubernetes.
An end-to-end sales performance dashboard built with Python and Plotly Dash, integrating PostgreSQL data to surface regional trends, product category insights, and executive KPIs in real time.