At JuniGadi.com, I built an end-to-end AI-powered automatic number plate recognition system using YOLOv8s and PARSeq. It achieved 98.04% exact-match OCR accuracy on Indian vehicle license plates.
I also developed 20 insurance document extraction pipelines that processed more than 1,000 policies, converting information from varied PDF templates into structured JSON.
At Codveda Technologies, I developed an XGBoost machine learning pipeline that achieved 95% accuracy and 0.92 ROC-AUC. I also containerized inference services with Docker and developed FastAPI REST APIs for deployment.
My research includes machine learning for hypertension screening using BCG signals and a Swin Transformer framework for plant disease detection. I’ve also built projects including ForgeGuard, an AI safety monitoring pipeline, and LeafCure, a plant health platform.
