Worked on integrating the InternLM2 model family into ONNX Runtime, focusing on both the microsoft/onnxruntime-extensions and microsoft/onnxruntime-genai repositories. Developed runtime recognition for the InternLM2Tokenizer, enabling dynamic model-tokenizer integration through updates to tokenizer configuration and dependency management using CMake. Implemented Python-based builders and model infrastructure changes to support InternLM2 models, including weight mapping and GroupQueryAttention handling for efficient CPU INT4 export and inference. Enhanced documentation and end-to-end validation improved deployment readiness and developer experience. Demonstrated skills in AI integration, C++, Python, and machine learning, delivering two features with a focus on robust model support.