Direct the analytics team's delivery, setting the metric definitions, pipeline architecture, and dashboard standards adopted across every team deliverable.

Built Pandas pipelines aggregating monthly CRMLS listing/sold data and merging in 30-yr mortgage rates.

Engineered key metrics (price/sq ft, close-to-list ratio, days on market) and applied IQR outlier flagging.

Designed Tableau dashboards on CA trends, ZIP comps, agent/office rankings, sales volume, units sold, and shares.

Selected for a 12-week program taking a production application (PropertyLite) from multi-tier networking through containerization, serverless, and CI/CD to a capstone deployment and SAA-C03 certification prep.

Evaluate AI-generated image edits across multiple dimensions at the pre-deployment training stage, producing comparative judgments that surface failure patterns and guide model improvements.

Ran SOTA vision-model training-data QA and evaluation against gold annotations to catch hallucinations and temporal errors.

Design weighted rubrics with calibrated scoring and penalty structures for each evaluation task.

Fact-verify model outputs against primary sources and flag IP and safety issues.

Contribute to additional AI training studies across video, audio, and text formats.

Built a live Streamlit dashboard over 119K transactions and 560K tickets ($11.4M across 10 seasons, 1,373 events) for pacing and sales-status decision support.

Redesigned the sales-pacing benchmark and backtested it against full history, correcting a structural bias that misranked events needing attention.

Cut dashboard load time from 3+ minutes to under a second by precomputing the analytics layer.

Quantified the post-COVID demand shift (~32% decline) to inform season planning and outreach.

Built a genetic-algorithm TSP router in Python for 20-25 daily stops, scoring routes on distance plus delivery-window penalties and emitting each day's ordered Google Maps route.

Automated data extraction from EKOS, the brewery's order-management system, straight into the routing pipeline, eliminating manual entry.

Clustered a year of wholesale accounts with K-means on invoice frequency and revenue, isolating a single-contract outlier and subclustering low performers.

Found 30% of accounts carried untapped growth potential while 25% of sales visits targeted limited-opportunity accounts; recommended reallocating visit capacity.

Built a 570K person-year analytic sample in Stata from ACS/IPUMS microdata (2009-2011, ages 24-39 with a bachelor's or higher), merging Sjoquist & Winters' STEM major classification and hand-verified state program data on strength and implementation year.

Estimated the effect of strong vs. weak merit-aid exposure on STEM degree selection using difference-in-differences and linear probability models across states and cohorts.

Lead operations for the club's data hackathon (~100 participants), owning logistics, timelines, and day-of execution so the event runs end-to-end.

Recruit and coordinate a judging panel of three industry professionals from established companies, serving as their single point of contact from outreach through awards.

Run the club's finances end-to-end: budget planning, sponsorship funds, and spend allocation across events and member initiatives.

Organized 20+ major events for ~100 cadets while enforcing standards and teamwork.

Supported squadron IT needs and day-to-day operational readiness.