Delaram focused on stabilizing the speculation testing pipeline for the EPFL-LAP/dynamatic repository, addressing issues that impacted the reliability of integration tests. She resolved MLIR syntax changes, corrected timing model paths, and aligned operator usage by updating the load5 operator to load2, which reduced test failures in subdiag and subdiag_fast tests. Her work involved targeted bug fixing and integration testing using MLIR and Python, ensuring that the speculation testing suite produced more consistent results. By decreasing flaky-test occurrences, Delaram improved the overall correctness of speculation-driven optimizations, demonstrating depth in debugging and maintaining complex testing infrastructure within a short timeframe.