Enhanced the model calculation by leveraging matrix calculation on GPU with PyTorch, ensuring scalable performance across grant datasets.

Built a memory-efficient pipeline, offloading data to disk with support for GPU acceleration, and utilized PSUtil to dynamically manage CPU/GPU memory.

Handled cross-platform compatibility development for devices, including compatibility between tablets and iPhones, as well as between different versions of the same device.

Used React Native and Figma as primary tools for front-end development.

Utilized Amplify as a network service provider and AWS for cloud computing services.

Deployed a block chain credit risk model based on XGBoost and TabNet

Employed PPS analysis and SMOTEENN techniques for account selection and feature engineering to ensure model accuracy and robustness.

Achieved 95\% target precision score and 0.9 overall F1 score in blind testing.

Enabled seamless API connection between frontend and backend within an AWS EC2 instance.