Nil Sagor

I'm a PhD candidate in Computer Science building applied AI systems from research prototypes through deployment. My work spans deep learning, graph neural networks, anomaly detection, and medical image analysis, with publications in Scientific Reports, IEEE conferences, and the Journal of Electronic Imaging.

I built an end-to-end Set Transformer anomaly-detection system that achieved 98.97% accuracy on synthetic data and 93% on CIFAR-100, including a Streamlit demo, FastAPI backend, Docker deployment, and GitHub Actions workflow. I've also developed dynamic GNNs for temporal link prediction and graph-based radiomics models for lung CT segmentation.