Worked on the ROCm/rocprofiler-compute repository to address a key bug affecting the accuracy of compute profiling for VALU workloads. Focused on refining the calculation of active threads by updating the PoP of VALU Active Threads metric to use a dynamic wave_size rather than a fixed value, ensuring the peak reflects actual hardware behavior. Adjusted normalization logic to represent the average number of active threads per wave, improving the fidelity of performance analysis. Utilized Python and YAML for system configuration and profiling logic, contributing to more reliable performance insights and supporting users in optimizing compute-intensive applications. Fixed one critical bug.