Abstract: The increasing volume of global visa applications has made manual assessment slow, inconsistent, and prone to human error, highlighting the need for intelligent and automated decision-support systems. This project presents a scalable Visa Approval Status Prediction System built using machine learning and integrated with a complete MLOps pipeline for continuous training, automated deployment, real- time inference, and monitoring. The system analyzes applicant features such as demographics, academic history, financial stability, work experience, and documentation quality using multiple ML algorithms including Logistic Regression, Random Forest, XGBoost, and SVM. The MLOps workflow incorporates GitHub Actions for CI/CD automation, Docker for containerized deployment, and AWS (EC2, S3) for cloud hosting and model registry. After preprocessing steps such as feature engineering, data balancing, and normalization, the final model delivers high accuracy with strong precision–recall performance, making it suitable for real-world visa decision support. The system ensures reliability, scalability, and adaptability through continuous monitoring and automated retraining, demonstrating an efficient and production-ready approach to modernizing visa evaluation processes.

Keywords: Visa Approval Prediction, Machine Learning, MLOps, CI/CD, Docker, AWS, XGBoost

Downloads: | DOI: 10.17148/IARJSET.2025.121247