Projects

A research-grade reinforcement learning system that trains a PPO agent to trade a 7-asset portfolio across tech, finance, energy, and ETF markets, combining NLP-driven sentiment analysis with rigorous walk-forward validation, multi-seed testing, and ablation studies to ensure scientific robustness and generalizability.

A Monte Carlo simulation framework in Python for modeling stochastic asset price dynamics under uncertainty. Asset price paths are generated using Geometric Brownian Motion under risk-neutral measure to estimate the fair value of financial derivatives, with sensitivity analysis across volatility and time-to-maturity parameters.

A tabular Q-learning agent implemented in PyTorch, trained on a Gridworld environment using Bellman optimality equations. The project explores policy learning under varying learning rates, evaluates performance via empirical return distributions, and includes hyperparameter sensitivity analysis.

A quantitative feasibility study modeling hydroelectric output under stochastic river-flow dynamics using flow-duration curves and HYDAT data. The analysis includes scenario and sensitivity analysis, tail-risk evaluation, and assessment of long-run economic viability.