Currently designated as a technology associate at Apnatech Technologies and Solutions, my role allows me to indulge in developing computer vision products as a primary responsibility. I am the lead developer in the PVIS (Photo vehicle Identification System) project in association with IRCTC (Indian Railway Catering and Tourism Corporation). We build an edge device which is capable of spatial analysis implementing Deep Learning based optical character recognition of RSID (Rolling Stock Identity) numbers on fast moving trains (120kmph) by recording and processing the video feed, from high resolution industrial cameras, in real-time.This year we plan to develop and deploy India’s first HANGING PART DETECTION system for rolling stocks ( Railway coaches ) with the capability to detect any misc hanging parts from the trains which can cause potential failure or harm to the regular railway operations.

Rules Removal | Document Cleaning Module• Developed a deep learning-based computer vision model to cater the rules removal in handwritten documents as a feature in Document Cleaning Module.• Prepared synthetic data of handwritten documents to meet the desired goal.• Performed multiple iterations of training and testing – to achieve the accuracy of more than 95% on specific cases.Annotation and Testing• Formed a sustained iterative data pipeline consisting of data preparation and annotation with input from 100+ freelancers and annotators.• Involved in testing the trained models on target specific scenarios, identifying error cases and suggesting improvement strategy.