Stefan Ruffini

• ›Own the platform tooling powering Fortnite's Discover and Shop teams, the systems they use to surface, rank, and experiment on what players see.
• ›Tripled those teams' experimentation throughput by building and scaling the A/B testing and experimentation capabilities they run on.
• ›Introduced standardized experimentation frameworks, including an experiment power calculator, cutting underpowered tests and improving result reliability.
• ›Drive prioritization and trade-offs across competing initiatives in a resource-constrained org.

• ›Owned experimentation ecosystem powering 200+ annual tests optimizing personalization and engagement.
• ›Designed data pipelines reducing analysis time by 90% and saving over $2M annually.
• ›Led near real-time performance data initiatives driving $50M+ monthly revenue and improving discovery efficiency by 10%.
• ›Established experimentation governance frameworks reducing data error rates by 5%.
• ›Mentored PMs and analysts, improving onboarding time and delivery consistency.

• ›Built real-time dashboards enabling predictive analysis and ad performance optimization across markets.

• ›Delivered consumer analytics improving media personalization and campaign optimization.
• ›Automated reporting workflows, cutting production time by 75%.
• ›Created dashboards that drove a 40% uplift in ad revenue.

A bandit-driven stock picker that runs a portfolio like an experimentation platform runs A/B tests. A risk-adjusted multi-armed bandit allocates capital across twelve strategy sleeves, each with its own mandate and risk budget, with a hard drawdown breaker underneath. Five-year backtests beat the S&P 500 on all 20 RNG seeds (+25.8pp median) with max drawdown 10 points shallower than the index (-14.5% vs -24.5%) and Sharpe 1.29 vs 0.80. Alt-data sleeves (insider flow, activist stakes, merger arb) validated on real EDGAR filings but held out of the live headline; the live daily paper-trading track runs the reproducible price-based core. Built in Python.