Enhanced the quality and accuracy of AI generated clothing by training and finetuning models on large-scale datasets. Spearheaded the development and launch of a portrait relighting and background replacement feature leveraging deep learning frameworks to enhance image realism and user experience.

Designed programming content and auto-gradable assignments for the Machine Learning and Applications course on the SAIL platform

Generated Digital Surface Models from multi-view stereo satellite images by optimizing NeRF for radiometric inconsistencies such as shadows, transient objects, and multi-date imagery. Utilized few-shot learning with Segment Anything Model (SAM) to enhance site-dependent building segmentation.

ES654 - Machine Learning

Instructor: Prof Nipun Batra