i-ScOper AI Enhancement
An ML-powered enhancement layer for SAP's Intelligent Test Scope Optimizer (i-ScOper) that intelligently prioritizes regression test cases using an ensemble of Random Forest, Gradient Boosting, XGBoost, and LightGBM models. The system dramatically reduces wasted testing cycles while maintaining full regression coverage for SAP enterprise deployments.
The Problem
SAP's i-ScOper identifies which tests could be affected by a code change, but it produces large, unprioritized test sets that teams run exhaustively. In fast-moving SAP development environments, running the full recommended set delays releases by hours. Teams needed a way to identify the highest-risk tests within the recommended set.
The Solution
A Bayesian-scored ML ensemble with rolling 4-fold cross-validation that learns from historical test execution data to assign risk scores to each test case. Tests are ranked by predicted failure probability, enabling teams to run the top-N% of tests first and catch 90%+ of failures in the first half of the test run. Includes a full EDA pipeline for data profiling and outlier detection.
Key Features
Challenges & Learnings
Test execution history data in SAP environments is sparse and imbalanced — most tests pass most of the time, making failure prediction a severe class imbalance problem. SMOTE oversampling combined with cost-sensitive learning was used to address this. Rolling cross-validation was critical to prevent data leakage from future test runs into training.
Business Impact
Reduces unnecessary regression cycles, accelerates release velocity, and cuts CI/CD pipeline runtimes for enterprise SAP application development. Deployed as a proof-of-concept for SAP's internal testing infrastructure team.
Future Direction
Integration with SAP's CI/CD pipeline for fully automated test prioritization, online learning to update models incrementally with each new test run, and extension to performance test prioritization.
Tech Stack
Quick Links
Want to see more of my work?
Back to Projects