I've researched black-box neural network security at DRDO's SAG Lab, recovering 95%+ of model parameters in under 10,000 queries and designing defenses that reduced attack success by about 40%.
I've built end-to-end systems ranging from a fraud-detection pipeline for 2.8M+ transactions that achieved 0.98 ROC-AUC to graph-based navigation, banking, social-feed, and version-controlled file-system projects. My work combines machine learning with efficient algorithms, custom data structures, and performance-focused engineering.
Alongside technical work, I lead cultural events as Cultural Secretary at Jwalamukhi Hostel and have represented my institute in stage and dramatics competitions.
