Systems architecture, data processing, and building scalable data platforms.
How I built a sophisticated prediction model that captures the complex relationships in NFL play-calling using Graph Neural Networks, multi-task learning, and quantile regression. The architectural decisions that made predicting football actually work.
A technical walkthrough of the Sports AI Assistant — a retrieval-augmented generation (RAG) system designed to handle complex, real-time sports queries using large language models, semantic retrieval, and live data integrations.
Real-world lessons and architectural decisions from building an NFL data pipeline that ingests from SportRadar to BigQuery. Why I chose Apache Beam over Spark, when serverless made sense, and how cloud-native architecture solved problems I didn't expect.
How I built a semantic entity resolution system using vector embeddings to transform natural language queries into structured data lookups, enabling OddsTrader's conversational AI to ground responses in real-time sports data.
I used to think scalability was the ultimate goal—that if a system could scale horizontally, it could handle anything. The Odds Importer taught me how data context can quietly destroy even the most perfectly designed distributed systems.
How I transformed a chaotic collection of hard-coded C# scripts into a scalable, observable data ingestion platform processing tens of millions of updates per day.