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{209953,
        author = {Dr MK Jayanthi Kannan and Mukul Mishra and Akshara Singh and Ananya Chaurasia and Akshara Srivastava and Akshat Srivastava and Nilabh Sharma},
        title = {MODELFORGE-AI @ AN INTELLIGENT MACHINE LEARNING CODE ANALYSIS, OPTIMIZATION AND DEPLOYMENT ASSISTANT},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {5},
        pages = {4853-4873},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=209953},
        abstract = {Machine learning applications increasingly depend on complex data-processing, model-training, and evaluation pipelines, making the maintainability and reliability of ML code an important engineering concern. However, existing development tools generally address individual aspects of the ML lifecycle, such as conventional code linting, hyperparameter optimization, model explainability, or deployment, without providing a unified understanding of the complete ML workflow. This paper presents ModelForge AI, a proposed intelligent platform for context-aware analysis, optimization, explainability, and deployment assistance for Python-based machine learning projects. The proposed approach uses Python Abstract Syntax Tree (AST) parsing to identify ML-related operations and construct a structured pipeline model representing stages such as data loading, preprocessing, feature engineering, data splitting, model training, and evaluation. A context-aware rule engine is then designed to identify potential ML-specific issues, including data leakage, inappropriate preprocessing order, missing validation, unsuitable evaluation practices, and reproducibility concerns. To improve the usability of detected findings, a Large Language Model (LLM), accessed through the Gemini API or locally using Ollama, is proposed to generate contextual explanations and improvement suggestions grounded in deterministic analysis results. The platform further incorporates optional Optuna-based hyperparameter optimization, SHAP-based model explainability, and automated generation of FastAPI and Docker deployment artefacts. The key contribution of the proposed framework is the integration of these capabilities through a shared ML pipeline representation, enabling a continuous workflow from code analysis to model improvement and deployment preparation. The current work focuses on system design, requirements, architecture, and conceptual methodology, while implementation and empirical evaluation are planned for the subsequent phase.},
        keywords = {Machine Learning Code Analysis, Context-Aware Static Analysis, ML Code Smells, Abstract Syntax Tree, Data Leakage Detection, Large Language Models, Hyperparameter Optimization, Explainable AI, ML Pipeline Analysis, Automated ML Deployment.},
        month = {October},
        }