Analyzed 100,000+ order records from Olist, Brazil’s largest online marketplace, to uncover the factors driving late deliveries. Using SQL for data extraction and Python (pandas, scikit-learn) for modeling, I built a logistic regression classifier that achieved 80.8% accuracy and raised precision by 26% after threshold optimization. The project focused on real-world tradeoffs between precision and recall, showing how small model adjustments can improve reliability for business operations. Summarized findings through visualizations and written conclusions highlighting actionable insights to reduce false alerts and improve delivery reliability.

Worked with Adsphere (a fictional social media ad platform dataset) to explore how advertisers structure and present campaigns. The goal was to determine whether advertisers with the most active campaigns and ads generate the most revenue, and to analyze how presentation choices such as formats and CTAs affect performance. Wrote over 10 SQL queries using INNER/LEFT JOINs, CTEs, and INTERSECT, and paired them with Python visualizations in Seaborn and Matplotlib to test correlations and highlight trends. The analysis revealed a positive correlation between ad volume and revenue (though not causal) and showed that CTA effectiveness is format-dependent.

Conducted a comprehensive analysis of U.S. ski resorts. Utilized Python and various libraries for data processing, statistical analysis, and visualization, assessing key metrics such as snowfall, terrain, amenities, and user reviews to derive data-driven insights.

Conducted informational interviews with Colorado Swim School management to gather qualitative insights and mapped findings using data journey and social world models. Collaborated with team members to synthesize data and perform secondary source analysis, evaluating social media presence and other external factors. Identified challenges, proposed design principles to improve information flow, and provided actionable recommendations for organizational improvement. Delivered findings in a professional presentation to Colorado Swim School stakeholders.

Designed and developed a dynamic, text-based two-player board game in C++ inspired by The Lion King, implementing interactive gameplay mechanics, file handling, and strategic decision-making features.

Determined the most profitable business and target location by using SQL to query Google reviews, then processed the data. Then surveyed potential customers to assess the competition.

Engineered end-to-end NLP pipeline integrating Spotify, Genius, and YouTube REST APIs to collect and store 57 songs and 5,000+ fan comments in a normalized SQLite database via automated ETL. Developed custom phonetic complexity metrics (rhyme density, syllable variance, stress variability) using NLTK and CMU dictionary; applied VADER sentiment analysis to quantify fan engagement patterns. Identified divergent lyrical complexity–popularity correlations (r = +0.43 vs −0.18) and visualized temporal trends across artist careers using Matplotlib.