Over a two-month period, contributed to the root-project/root and ferdymercury/root repositories by developing and stabilizing deep learning features in C++. Focused on improving reliability and maintainability, addressed a critical infinite loop in the ReduceMean operation and enhanced code consistency through clang-format compliance. In ferdymercury/root, implemented regression tests for tensor operations and delivered targeted bug fixes for TMVA Neural Network Ops, including robust ONNX parser input validation and accurate pooling shape inference. Leveraged skills in C++ programming, algorithm optimization, and unit testing to ensure production reliability, with each change accompanied by tests to reinforce correctness and support downstream deployments.