Redwanur Rahman

I've built full-stack MERN platforms and production-grade deep learning pipelines while completing my Computer Science and Engineering degree at BRAC University.

For my final-year thesis, I designed a real-time throat pathology detection pipeline that achieved 92.59% accuracy across eight diagnostic categories. I trained a multi-teacher knowledge distillation architecture using EfficientNet-B4, Swin Transformer, and ConvNeXt-B to create a lightweight EfficientNet-B0 model for consumer-grade GPU inference.

I've also built Urban Fix, a community reporting platform with emergency alerts, geolocation, identity verification, and role-based workflows, and CarKoi, a vehicle buying and rental marketplace with pricing, billing, recommendation, and inventory logic.

At BRACU Duburi / DuboTech Digital Ltd, I lead official website development and contribute across computer vision, robotics, software engineering, and automation projects. I enjoy moving quickly between product engineering and applied AI work, from React and Node.js applications to PyTorch-based ML systems.