Worked on stabilizing the intel/vpl-gpu-rt repository by addressing resource management and error handling issues in GPU-accelerated video processing pipelines. Focused on fixing destructor logic across several components to resolve memory leaks and prevent uncaught exceptions, thereby reducing crash surfaces in production environments. Applied C++ skills to enhance memory management, wrapping destructor bodies with try-catch blocks to mitigate process termination from exceptions such as std::bad_alloc. Improved teardown safety by ensuring proper release of loaded library handles before device deletion on error paths. Addressed static analysis findings, specifically Coverity issues, to strengthen overall reliability and memory safety in the codebase.