Worked on enhancing build reproducibility and caching for Intel-tensorflow/tensorflow and Intel-tensorflow/xla by improving HLO module fingerprinting. Developed custom field support within the fingerprinting process, making fingerprints more granular and attribute-driven to enable deterministic builds across different environments. Coordinated changes across both repositories to ensure consistent fingerprint semantics, updating core C++ components such as HloModule::Print and HloModule::ToFingerprint. Focused on feature development rather than bug fixes, the work leveraged skills in build systems, compiler development, and low-level programming. These enhancements reduced unnecessary rebuilds and established a more robust foundation for optimization and deployment workflows in C++ projects.