Alexander Sannikov enhanced the intel/pti-gpu repository by implementing device-initiated MPI communications profiling within the Unitrace tool. He introduced new logging capabilities that capture detailed performance data for device-initiated MPI events, including their sequence numbers, thereby improving observability of GPU-driven MPI paths. Working primarily in C and C++, Alexander applied his expertise in low-level systems programming, MPI, and performance profiling to instrument the profiling layer and update related documentation. This work enables faster diagnosis and optimization of GPU-enabled MPI workloads in enterprise environments, reflecting a focused and technically deep contribution over the course of the month without reported bug fixes.