At WSP India, I developed a CatBoost surrogate model to predict building energy intensity and thermal comfort, then applied NSGA-II to compare design alternatives across energy use, comfort, and capital cost. I also built an interactive Streamlit application for reviewing optimized designs and their trade-offs.
I’ve applied machine learning and time-series methods to energy forecasting and BIM clash prioritization, including a system that reduced manual coordination review effort by approximately 40%. Earlier, at Enphase Energy, I designed and reviewed solar PV systems and built prototype generation-forecasting and anomaly-detection models.
