Segmented 2.2 million trades at Monex into four trader personas and found that two of them place 92.6% of all tickets, which changes who the retention program should be built for.

A simulated onboarding experiment built in Python. Adding a progress bar lifted completion from 52.8% to 61.8%, and almost all of the gain came from new users.

A web app I built in PHP and SQL that scores trading customers on churn risk and pairs every score with an action for the retention team.

Classified more than 12,000 Uber app reviews by feature area and sentiment. The ratings turned out polarized rather than mediocre, and payment issues led every complaint category.

A Monte Carlo and Solver model for Vail Mountain, built in Excel with Crystal Ball. Across 1,000 simulated seasons it returns a mean profit of $33.7M and says stay closed through October and November.

A COVID relief organization I founded in Jakarta. More than 1,000 care packages distributed to families, funded through six corporate partnerships.

Six LSTM models trained on 290 NBA players, one per statistical category, that rank the ten best season lineups from a user's roster. Built in Python with Selenium, BeautifulSoup, and SQL.

An AI scent assistant proposed for a New York custom perfumery, with a six step customer flow, a working chatbot demo, and a four phase rollout plan.

An industry and company analysis of Texas Instruments, the analog semiconductor leader holding 19% of an $89.55B market, built as a stakeholder deck.

The capstone case study for the Google Data Analytics Professional Certificate, which I finished in July 2026.