On the DevoGraph DHGNN project, I built data pipelines and sparse hypergraph representations from cell-tracking and connectome data, then implemented and trained a dynamic neural network for cell-fate prediction. The model reached 0.98 accuracy, and its learned embeddings showed improved structure over raw features.
I’ve also developed AI applications spanning climate research, sign language recognition and translation, and real-time computer vision. My work includes a LangGraph climate analytics pipeline and a sign language classification application that achieved 99.52% accuracy.
