Vamsi Manchala contributed to both the google-ai-edge/ai-edge-torch and tensorflow/tensorflow repositories, focusing on cross-framework deep learning and compiler infrastructure. He unified interpolation behavior between JAX and PyTorch by standardizing NCHW attribute handling, which improved consistency and reduced integration friction. In addition, he stabilized CI pipelines by re-enabling tests and updating TensorFlow dependencies. For TensorFlow, Vamsi developed an MLIR dialect registry for TensorFlow Lite, enabling extensibility of MLIR-based representations, and streamlined optimization tooling by removing unused passes and dependencies. His work, primarily in C++ and Python, demonstrated depth in build optimization, dependency management, and cross-framework machine learning engineering.