denoslab/starfish-fl

Built on a distributed architecture with a Router coordination hub, site-based Controllers, and a Typer-powered CLI agent, Starfish-FL implements 20+ federated ML tasks spanning regression, survival analysis, and count data models in both Python and R. It integrates biostatistical methods (logistic regression, Cox models, MICE imputation) with proper inverse-variance meta-analysis and diagnostics, while supporting optional LLM agents for adaptive aggregation and autonomous experiment orchestration. Docker-based deployment enables privacy-preserving collaborative analytics across healthcare, finance, and carbon-aware computing domains without centralizing sensitive data.