Scalable Validation and Continuous Verification of AI/ML Systems on AWS Using Python-Based Automation

Abstract

Data drift, model degradation, and inference inconsistency are the main issues related to the operational implementation of AI/ML systems in cloud-native settings. This paper suggests a Python-related code of scalable validation and the constant verification of AI/ML models on AWS. It incorporates automated validation of the dataset, performance benchmarking and drift detection in the containerized deployments of AWS EKS, Lambda and S3. Datapoints, e.g., precision, recall and distribution variance are introduced into workflow CI/CD pipelines as real-time tools. The empirical assessment exhibits better reproducibility, few post-deployment errors and better traceability. The methodology sustains the trustful, automatic and undisrupted AI/ML usage in series cloud environments that can be scaled.