At Medal Lab, IIT Bombay, I develop deep learning models for medical imaging and support clinical digital pathology deployment. For cancer cell segmentation and grading, I worked with datasets from ICMR - National Institute of Pathology, New Delhi, and achieved a Dice score above 0.85 and an F1 score above 0.6 across different centres.
At TMH - Tata Memorial Hospital, I worked on a deep learning model for cosmetic evaluation of breast cancer conservative treatment, improving relative accuracy by 3.1% after fine-tuning on 1,000 images. My projects also include Contrastive-BYOL, which improved accuracy by 3% to 12% over SimCLR and BYOL baselines, and AI applications using retrieval-augmented generation, LangChain, and LLaMA models.

