❯cat about.confos arch linux x86_64wm dwm (patched)shell zsheditor whatever workscgpa 9.24papers 1 (ieee)status —# sleep.sh not found# todo: fix the overflow❯
scroll
// research
PUBLISHED WORK
ieee xplore • nov 2025
Semi-Supervised Oral Cancer Image Classification Using GAN-based Augmentation and EfficientNet-B3
Sayan Pal, Utkarsh, Shoham Chakraborty, Saurabh Bilgaiyan, Chiranjib Parida 2025 3rd DMIHER International Conference on AI in Healthcare, Education and Industry Wardha, India • DOI: 10.1109/IDICAIHEI65991.2025.11379848
A semi-supervised framework for oral cancer detection combining EfficientNet-B3 with GAN-based data augmentation to tackle limited annotated medical data. Integrates supervised classification, semi-supervised feature learning via SSGAN, and few-shot evaluation using Prototypical Networks — towards scalable AI-assisted cancer screening.
Full mail stack on Debian Trixie. Postfix, Dovecot, TLS, Fail2Ban, IPv6. Built it because paying for email is a philosophical failure. Production-grade security, zero secrets in version control.
OpenEMR research on patient data flow in Indian hospitals. Deployed on GCP, OAuth 2.0, REST APIs, custom module. The real value was navigating a production-grade open source codebase and understanding healthcare infrastructure end-to-end.