Projects

Building-level dataset integrating three City of Boston public datasets to infer heating systems, paired with a client-facing property map that helps BCAN identify high-emission buildings for decarbonization outreach.

End-to-end data pipeline using the MA Legislature Public API to collect and structure 10 years of petition data. Built a refiling fatigue analysis dashboard to quantify repeated filings and their impact on success rates.

LLM-powered article analyzer exploring how prompt design influences model outputs. Demonstrates the impact of prompt engineering on summarization quality and response variability.

Python/TensorFlow pipeline that transforms raw audio into mel-spectrograms for CNN-based emotion classification. Fine-tuned model performance through structured hyperparameter sweeps across speech and song datasets.

Graph analysis of the US interstate highway network in Rust, using degree and closeness centrality to rank states by their strategic importance in the national road system.

Statistical and machine learning analysis of delivery performance data, identifying key correlations between personnel performance, delivery times, and customer satisfaction.