Worked on enhancing profiling capabilities in the pytorch/pytorch repository by delivering an upgrade to the Strobelight GPUEvent Profiler. Focused on enabling argument value and type recording for ATEN operators, the work involved updating the Dispatcher to accept additional parameters for argument addresses and types within profiling calls. This approach allowed for more granular and precise performance analysis of GPU workloads during operator execution. The implementation was carried out using C++ and leveraged expertise in GPU programming and profiling techniques. The contribution improved the depth of instrumentation available to developers, supporting more detailed investigation of performance characteristics in PyTorch.