Shell-Research

My internship at Shell India provided a high-level perspective on smart grid automation, where I collaborated with international teams to analyze the software architecture of grid management systems. I conducted foundational research on critical components like Network Topology Processors (NTP) and State Estimation (SE) , analyzing Linux Foundation software stacks like SOGNO and prototyping their deployment on Kubernetes. This experience gave me a clear understanding of the data-driven challenges in grid management and inspired me to build a core component of such a system myself. Building on this, I undertook an undergraduate research project with Prof. Shanti Swarup to develop a complete microservice-based probabilistic load forecasting (PLF) framework. I designed and implemented an LSTM-based sequence-to-sequence model and rigorously optimized it using Optuna for hyperparameter tuning, which improved the model's Mean Absolute Error by 5.7% (from 0.0348 to 0.0328). To move beyond a black-box solution and create a system suitable for real-world grid automation, I integrated SHAP to provide full model explainability, delivering transparent forecasts with 95% confidence intervals.