Amalur: when data integration meets machine learning

Training data rarely lives in one place. It is scattered across tables, files, and silos, each with its own schema and quirks. Amalur asks a simple question: what if data integration and machine learning were designed together, rather than bolted on after the fact?

By treating integration as part of the learning pipeline, Amalur automates much of the tedious work of bringing scattered sources into a form that models can actually use — the first step toward our broader Model Lake vision.