Article Published In Vol.9 (March-April 2021)

AI-Driven Software Testing and Development: Enhancing Automation, Efficiency, and Reliability in Agile and DevOps Environments

Author : Sathiyendran Ganesan and G. Arulkumaran

Increased levels of complex integration and continuous delivery (CI/CD) pipeline sophistication within current Agile and DevOps systems mandate advanced AI-facilitated solutions to run automated tests of software. Inefficiencies inherent in legacy test frameworks’ capabilities to respond to the fast-tracked rate at which software changes necessitate accepting intelligent test measures that improve flaw detection, outlier identification, and proactive maintenance. This work proposes a hybrid AI testing framework that integrates Bidirectional Encoder Representations from Transformers (BERT) and Long Short-Term Memory (LSTM) networks to optimize log analysis, anomaly detection, and failure prediction in CI/CD processes. The architecture begins with gathering data from CI/CD pipeline logs, extracting critical metadata such as run status, error messages, and pipeline performance metrics. The missing values are handled during the preprocessing process through imputation techniques, and the logs are structured by using regular expressions (Regex) and Term Frequency-Inverse Document Frequency (TF-IDF) for efficient analysis. BERT is employed to get context information from log messages through utilizing its deep language understanding ability. LSTM simultaneously processes log data in sequences to detect long-term dependencies and pipeline execution anomaly prediction. Both BERT and LSTM combination also improves prediction precision (92-97%), optimizing computation cost with improved precision, recall, and F1-score, outperforming individual models. This machine learning-based framework improves execution time, defect discovery, and the consumption of resources, promoting intelligent and adaptive test automation. The research achieves considerable improvement in software dependability, DevOps measurements, and anticipatory system maintenance, leading the way to intelligent, elastic, and AI-based software testing methodologies.

Keywords: AI-driven testing, BERT, LSTM, CI/CD pipeline, anomaly detection, log analysis, DevOps automation, predictive maintenance, software reliability, hybrid AI model