Contributed to the csc-training/summerschool and Lumi-supercomputer/lumi-userguide repositories by developing GPU debugging infrastructure, optimizing CUDA kernel exercises, and enhancing technical documentation. Leveraged C++ and CUDA to implement thread-level output, error handling, and performance improvements for hands-on curriculum modules, addressing issues such as enum misuse and include path errors to stabilize reductions exercises. Expanded educational content with new optimization demonstrations and clarified learning objectives to improve student comprehension. Additionally, revised Singularity container documentation to clarify PRoot-based build options, streamlining onboarding and deployment guidance. Emphasized maintainability and readability throughout, applying technical writing and containerization expertise to support both users and developers.