Worked on the viamrobotics/viam-cpp-sdk repository to enhance the robustness of machine learning model serving by addressing input tensor validation within the MLModelService. Focused on refining the logic to correctly handle both single-tensor and multi-tensor input scenarios, the work ensured that single-tensor cases bypass unnecessary name checks while multi-tensor cases strictly require all expected tensors with correct data types. This C++ and gRPC-based solution centralized validation and error handling, reducing runtime errors from misconfigured inputs. The update improved the reliability of ML inference pipelines and provided clearer error messages, contributing to a more predictable developer experience and safer model deployment.