Notes from turning agent workflows, personalization loops, AAmplify, FlumeWorks, and AI transformation ideas into software people can actually use. View all writing
2026Thompson Sampling: how explore meets exploitA Bayesian bandit algorithm that solves the exploration-exploitation tradeoff. How it works, why it's elegant, and where it shows up in real systems.30/0310 min readGeometric mean vs Z-scoreUnderstanding why we moved action group scoring from geometric mean to z-score normalization.17/0212 min readFrom DS notebooks to realtime pipelines: diversity scores at scaleHow we took a data science scoring algorithm and rebuilt it for realtime distributed systems.15/0110 min read
2025Audience, triggers, campaign, campaign_allocationsHow audience eligibility, triggers, and campaign allocations work together in the learning pipeline.17/1112 min readVariants, alternates, labels, componentBreaking down the Composer data model: what variants, alternates, labels, and components actually mean.17/115 min read