Yasser -

• Developed ML transaction matching solution for Oracle Cloud EPM using adaptive two-stage pipeline (rule-based filtering + classification), implementing continuous learning from automated labeling system with in-database inference for production-scale deployment. • Designed the integration for Oracle EPM, defining how to store data science artifacts, creating supporting database tables, and establishing workflows to automate retraining and deployment. • Project showcased at KScope25, a premier international Oracle conference highlighting real-world enterprise AI solutions.

• Researched code search approaches for license compliance detection in AI-generated code, evaluating multiple similarity detection methods including text-based, token-based, AST-based, and embedding-based approaches across precision, recall, and scalability dimensions. • Prototyped two-phase retrieval pipeline combining OpenSearch for textual similarity with code embeddings for semantic classification to detect Type-3 and Type-4 code clones at scale. • Conducted infrastructure performance analysis comparing search solutions, benchmarking indexing throughput, query latency, and resource requirements to inform architectural recommendations for enterprise-scale deployment.