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{201371,
        author = {Dimpal Bhangale and Shruti Bonde and Deliya Rane and Dr. Jagadish Jakati},
        title = {ModelOps: A Lightweight End-to-End Machine Learning Pipeline for Bank Customer Churn Prediction with Multi-Model Comparative Evaluation},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {12},
        pages = {3474-3480},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=201371},
        abstract = {This paper introduces ModelOps, a lean, deployment-oriented Machine Learning Operations (MLOps) framework built for real-time bank customer churn prediction. The system covers the full ML lifecycle from raw data preprocessing and comparative model training, through evaluation, serialization, API-based inference, containerization, and cloud deployment, all the way to CI/CD integration unified within a single coherent and reproducible pipeline. Three binary classifiers are trained and benchmarked against a feature-reduced subset of the Bank Churn dataset: Logistic Regression, Decision Tree, and Random Forest. Cross-metric evaluation spanning accuracy, precision, recall, and F1-score shows that Random Forest delivers the strongest predictive results (F1-score: 0.82), while Logistic Regression holds its own with considerably less computational cost. A key academic contribution of this work lies in the deliberate design and empirical validation of a resource-constrained MLOps architecture one that places operational simplicity front and center without cutting corners on functional completeness. The production-ready model is exposed through a RESTful FastAPI endpoint, packaged inside a Docker container, and deployed on AWS EC2, with its frontend counterpart served from AWS S3. Automated CI/CD via GitHub Actions handles continuous retraining and redeployment whenever the repository is updated. Together, these components demonstrate the real-world viability of full-stack ML deployment in low-resource settings and offer a reproducible reference architecture for the MLOps community.},
        keywords = {MLOps, Customer Churn Prediction, Logistic Regression, Decision Tree, Random Forest, FastAPI, Docker, AWS, CI/CD, Machine Learning Pipeline.},
        month = {May},
        }