Risk-aware devops: leveraging ai and ml algorithms for adaptive software delivery

The growth in the demand for faster software delivers influence the DevOps pipelines to keep up with that demand. On the otherhand, the risk management is falling apart if the pipelines are not properly designed. This study presents a Risk-Aware DevOps framework, which contains predictive and adaptive risk mitigation workflows using artificial intelligence (AI) and machine learning (ML). The proposed framework contains three parts: risk prediction/fusion, adaptive orchestration, and feedback-driven learning. Primarily this framework uses machine learning to predict failures, rank risks and invoke mitigation strategies automatically. In this study, a publicly available NASA PROMISE JM1 dataset was employed for predicting software bugs using the proposed framework. ML models such as Logistic Regression and Random Forest were trained and tested on NASA PROMISE JM1 dataset for prediction accuracies and both attained (ROC-AUC ≈ 0.70). We have also simulated a gating policy which exhibited improvements in operations, lowering failure rates and mean time to recovery while maintaining automationat an acceptable level. The results obtained in this study show that it is possible to employ ML-based decision pipelines to lower the risk in software delivery pipelines.All of the datasets, experimental scripts, and trained models used in this study can be found on GitHub at https://github.com/rajeevkankanala11/RiskAwareDevOps-AI.