Shivendra S.

Applied Scientist @Bosch India

Bengaluru, IN

Rare scenario image generation using GAN and diffusion models Enhanced object detection models (Traffic light detection, Lane detection - ADAS) by generating new images for rare classes using domain adaptation. Deployed the enhanced model into the product pipeline, resulting in a 10% improvement in class-wise model Intersection over Union (IOU), saving 2-3 weeks of data collection time and reducing costs by approximately per domain.Semantic similarity detection model on textual requirements data. Extracted relevant manufacturing requirements from past projects based on new stakeholder requirements, reducing the average processing time from 20-25 mins to 5-8 mins per requirement. Conducted data analysis and generated benchmark data using candidate pooling. Deployed BERT model into a web app, successfully handling 50+ users reducing man-hours required by 70%.

Dr VSEC

High School, Computer Science

Indian Institute of Technology, Delhi

Bachelor’s Degree, Chemical Engineering,