On E-Doctor, I built a multi-role healthcare backend with TypeScript, Express.js, Prisma ORM, and PostgreSQL. I used Redis caching alongside RAG-based medical assistance to bring clinical response retrieval latency under 200ms.
For ShasthoAI, my final-year project, I developed a YOLOv8 fracture-detection pipeline and improved model mAP from 0.64 to 0.75 over baseline benchmarks. I deployed the model as a Python FastAPI microservice and integrated it into a full-stack diagnostic workflow.
At the Network and IoT Lab at United International University, I researched hardware-software IoT integrations, engineered departmental prototypes, and structured open-source IoT repositories. I also mentored undergraduate students in lab sessions on network protocols, microcontrollers, and system architecture.
