Led machine learning teams and built large-scale AI systems across computer vision, multimodal learning, and LLM applications, delivering ML models in production at scale, used by multiple Fortune 500 enterprise customers.
Developed perception algorithms for autonomous mobile robots, enabling robots to interpret sensor data to understand outdoor environments, supporting localization and reliable navigation.
My PhD, advised by Ram Nevatia, focused on adapting offline-trained object detectors to new video datasets. I proposed novel approaches for unsupervised online adaptation, including incremental learning and multi-instance learning frameworks to improve object detection accuracy leveraging unlabeled data.
Conducted undergraduate research on recognition of scanned and handwritten Devanagari text, developing methods to clean and straighten scanned documents and recognize handwritten numerals using machine-learning classifiers including MLP, LVQ, and KNN.
Selected publications. Full list available on Google Scholar.
