At WnCC IIT Bombay, I architected an American Put Option pricing engine using Binomial Trees, neural networks, and reinforcement learning. Its deep neural network achieved a 0.0299 mean absolute error, and I evaluated a Double DQN agent across 5,000 Monte Carlo episodes.
Across my independent projects, I built machine-learning systems for supply-chain delivery risk, log anomaly detection, customer churn, industrial defect detection, and patient readmission analysis. My work includes a tuned XGBoost model with 0.987 ROC-AUC and a GRU log anomaly detector with 99.98% accuracy.
