Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{205453,
        author = {Keerthana K and Ms.E Uva Shakthi},
        title = {AUTOMATED FEATURE STORE AND METADATA ENGINE FOR ML EXPERIMENTS},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {1},
        pages = {6754-6759},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=205453},
        abstract = {MLOps environments are plagued with critical inefficiencies, such as disjointed feature stores, irreproducible experiments, and failing in production due to training-serving skew. Data scientists waste time reinventing features. Enterprises lose model accuracy due to undiscovered feature drift. In this paper, we present Automated Feature Store and Metadata Engine for ML Experiments, a unified solution tightly integrating feature engineering, experiment tracking, metadata lineage, and monitoring into one comprehensive system. Our proposed architecture encompasses Feast as the feature store, MLflow for experiment tracking, Apache Spark as a scalable compute engine to build features, Redis for online feature serving, and Evidently AI for drift detection and alerts. Our solution automates feature versioning, lineage, and reproducibility while providing flexibility of batch and real-time feature pipelines. Through our experimentation, we show improved experiment reproducibility, decreased time-to-experiment, and online feature serving at millisecond latencies. This framework can scale to support enterprise-level MLOps workflows with a one-stop-shop solution ready for production.},
        keywords = {Feature Store, MLOps, Feast, MLflow, Metadata Management, Drift Detection, Redis, Apache Spark},
        month = {June},
        }