• Work on SEEN function rebuild at IMDb, which allows users to mark a movie as watched with explicit and implicit marking(check-in, review & rating). The rebuild version will be compatible with current IMDb modernization architecture and updated business logic.• Create an event driven component with AWS Lambda using Java that will automatically process implicit actions and adds them to the Seen List in DynamoDB.• Using amazon internal tools to set up the pipeline and all the required resources stack using AWS CDK. The final submission will also include error handeling, UT/ST, metrics and logging.

• Analyzed current operator design patterns in TensorFlow; Prototyped the Python frontend APIs of Dilation and its two backpropagation operators with the MindSpore AI standard.• Developed the operators’ C++ backend with native kernel class; Supported computations with 12 different precisions of tensors; Added support for dynamic shape input besides exsisting static shape input.• Ensured the relative error <0.002 vs. NumPy for all computations by converting intermediate data type to fp32 and transforming math formulas to avoid numerical instability; Reduced for loops to minimize branch divergence.• Performed unit tests with 100% coverage and system tests with PyTest; Wrote design, delivery and API docs.

• Focused on the problem of data sparsity in image-based particle recognition methods; represented particle jets images as graphs and explored various channel combinations as node features; determined the most appropriate representation strategies.• Defined edges in a static or dynamic manner and computed connectivity in static graphs using KNN or radius neighbors.• Built various GNN model architectures for end-to-end tau particle recognition using current cutting-edge methods in graph deep learning research, including graph convolution, graph SAGE, graph attention, and dynamic edge convolution.• Analyzed model performance from multiple perspectives and provided possible explanations. Benchmarked model inference on GPUs and provided guidance for users through documentation and code.• Developed CLI tools in Python to automate training and inference, improving development efficiency by 40%.