OM SURASE

Developed a B2B generative AI service that automates candidate profile scrubbing and shortlisting, achieving a 100% reduction in sourcing time and currently utilized by 5+ organizations.

Delivered an end-to-end AI resume analyzer feature using Next.js, Flask, and MongoDB, significantly enhancing recruitment efficiency with a 75% candidate-screening accuracy for 250+ users.

Applied advanced technologies including Python 3, Redis, Celery, Next.js, Flask, Redux Toolkit, TypeScript, MongoDB, PostgreSQL, and Gen AI to build robust, scalable solutions.

Engineered an image segmentation solution using PyTorch & RCNN, increasing irregular field growth detection accuracy by 30% and reducing manual field inspection time by 40%.

Developed a comprehensive dataset by augmenting 150+ images with 6 transformation techniques, improving model accuracy by 22% over baseline implementations.

Raised F1-score for growth detection to 92% and reduced false positives by 25%, demonstrating strong analytical and model optimization capabilities.

Designed a triple-stacked LSTM neural network architecture for De Novo drug generation, successfully producing 50+ novel chemical formulas with 83% structural validity for potential psychotic applications.

Employed Python 3, PyTorch, Matplotlib, BeautifulSoup, Selenium, and Linux for chemical formula generation and analysis.