Worked on the NVIDIA/cuda-quantum repository to address a critical issue in quantum computing workflows, specifically improving the correctness of sampling in mixed-basis measurements. The solution involved reworking gate operation logic in C++ and Python so that sampling flushes only occurred for qubits that had already been measured, thereby preventing truncation of multi-qubit bitstrings and preserving mid-circuit measurement semantics. Regression tests were added in both languages to verify that mixed-basis kernels returned correctly-sized bitstrings with proper qubit ordering. This targeted bug fix enhanced the reliability of quantum sampling and was accompanied by updated documentation and comprehensive test coverage.