State-adaptive policy regularization (SAPPS) for smooth control in continuous-control reinforcement learning, with simulation and real-world experiments.

Physics-driven hydroelectric dam RL environment — Torricelli hydraulics, diurnal pricing & monsoon surge. Built for Meta × Scaler OpenEnv Hackathon.

Wavefront-sensorless adaptive optics using action-regularized reinforcement learning for optical satellite communication, with simulation environments and reproducible experiments accompanying the Optica Open preprint.