Sauhard Bhatt

Anomaly Detection in Pipeline Metrics: Employing AI to Detect Anomalies in CI/CD Pipeline Metrics

Abstract:

Continuous Integration and Continuous Deployment (CI/CD) pipelines are critical to the current state of software development, enabling pace and quality. Nevertheless, the larger these pipelines are, the harder it is for human operators to detect performance decline. In this study, the researcher analyses how artificial intelligence can automate anomaly detection in pipeline metrics, namely build duration, test execution time, and deployment frequency. This paper will use an Isolation Forest algorithm and Long Short-Term Memory networks to identify patterns in underlying system problems from a dataset of 429 pipeline execution logs. The tools used include machine learning libraries (Python) such as Scikit-learn and TensorFlow, along with Prometheus for metric gathering and Grafana for visualization. The findings show that AI-based monitoring provides proximal insights into the detection of bottlenecks and infrastructure instability, which traditional threshold-based notifications cannot. This research will provide a framework for shifting reactive troubleshooting to predictive maintenance in the DevOps setting, maintaining delivery cycles as codebases grow increasingly complex.

Profile:

Sauhard Bhatt is an accomplished Cloud and DevOps engineering leader with over 19 years of experience architecting, modernizing, and governing enterprise-scale hybrid platforms across healthcare, financial services, and highly regulated environments. He currently leads large-scale cloud transformation initiatives, shaping organizational platform strategy, operational excellence, and security posture. Core expertise spans Azure governance, DevOps automation, identity and access modernization, and middleware transformation across WebLogic, WebSphere, container, and Kubernetes-based platforms.

In his current role, he provides technical and architectural leadership for enterprise-wide initiatives, including Azure Arc–based hybrid governance, standardized CI/CD platforms using GitHub Actions, infrastructure-as-code adoption, and adoption of AI-assisted engineering practices by integrating Microsoft Copilot into development and DevOps workflows. In parallel, he researches and evaluates Internal Developer Platform (IDP) capabilities, focusing on self-service provisioning and failure-proof automation to accelerate delivery while maintaining governance and security controls.