Hudson Pagni
I build data systems that turn raw numbers into clear decisions, from machine learning simulations to analytics that run at scale. I am working toward a career in data analytics and data engineering.
A little about me
- Education
- B.S. Statistics & Data Science, UCLA
- Graduating
- June 2028
- Focus
- Data Analytics, Sports Analytics & Data Engineering
- Based in
- Los Angeles, California
I study Statistics and Data Science at UCLA, graduating in June 2028, and I am aiming at sports analytics and data engineering: building the models and the pipelines that turn raw data into decisions someone can act on. Most of my recent work sits there. A cross-sport fantasy platform that had to make a 162-game baseball season and a 12-game football season comparable, and a Counter-Strike 2 system that trains player-specific behavioral models from professional match demos, simulates matches between them, and searches tens of thousands of lineups for the five that perform best together. I grew up in San Diego and have been reading box scores far longer than I have been writing code, which is roughly how the statistics interest started. When I am not working on projects or coursework, I am a diehard Padres fan.
Featured projects
Gorilla Draft: Fantasy Leagues That Draft Teams, Not Players
A live fantasy platform where managers draft whole clubs across ten team sports, built around a published scoring card that prices a baseball win and an NFL win onto the same 1000 point season, and shipped from one React bundle to both the website and an iOS build.
View project
Counter-Strike 2 Dream Team: Cloning the Pros to Search for the Best Lineup
A machine learning system that trains player-specific behavioral models from professional match demos, tests whether they stay distinguishable on unseen maps, and simulates 37,032 candidate lineups on a GPU to find which five perform best together.
View project
rsi2bot: Testing RSI(2) Mean Reversion Behind a Fail-Closed Trading Agent
A paper-by-default equities agent for the official Robinhood Agentic Trading MCP server. The backtester imports the same strategy, sizing, and risk modules as the live path, and under the tested universe and period RSI(2) mean reversion roughly matched buy and hold.
View project
Multi-Agent Roblox Game Studio
Two Roblox games built as MVP builds through a development system I designed: specs of record, frozen inter-module contracts, scoped agent tasks, and headless verification gates that tested generated changes before integration. The flagship is about 57,000 lines of Luau across 109 modules.
View project
Instagram Automation and Analytics Platform
A distributed platform that operates a fleet of isolated Instagram sessions from one place, generates their content with AI, and tracks engagement per account.
View project
RVC v2 Voice Model Training Pipeline
A training and release pipeline built on the RVC v2 framework, used to train and publicly release more than ten voice models.
View project