Replication Package: Architectural Design Decisions for Managing Features, Metadata, and Models in Machine Learning Platforms

This repository contains the replication package of the paper "Architectural Design Decisions for Managing Features, Metadata, and Models in Machine Learning Platforms" submitted to 19th IEEE International Conference on Software Architecture (ICSA 2026). The study proposes architectural design decisions (ADDs) and options for feature, metadata, and model store in ML platforms. First, we conducted a gray literature review to build the framework, and second, we presented it to experts forvalidation. We identified 15 ADDs, 76 options, and 40 decision drivers. This artifact package contains: Gray Literature Review (GLR): A folder with the GLR process, the selected sources, and the coding of the selected sources as Atlas.ti project Expert Interviews (EI): This folder contains the interview protocol, the anonymized interview transcripts, and the coding of the interview transcripts as Atlas.ti project NB: For more detailed information on the artifact, read the README.md file