Rapid and reliable diagnostics of lithium-ion batteries are critical for electric vehicles and stationary energy storage systems. As the number of batteries in operation continues to grow, scalable methods are needed to assess battery health, degradation mechanisms, and remaining useful life under heterogeneous usage conditions. In this project, we develop a pulse-based battery diagnostics framework combined with machine learning to enable rapid health estimation of lithium-ion batteries. Short electrical pulses are applied to batteries and the resulting voltage responses are analyzed to extract electrochemical signatures related to internal resistance, diffusion processes, and polarization dynamics. Large datasets of pulse responses from retired commercial batteries will be processed to extract diagnostic features. These features will be used to train machine learning models capable of predicting battery state-of-health and identifying degradation patterns across different chemistries and capacities. To improve model robustness under data scarcity, physics-based battery simulations will be used to generate synthetic training data. These simulations will be conducted using the PyBaMM battery modeling framework. The computational workflow involves large-scale time-series processing, machine learning model training, and physics-based simulations. GPU resources are required to accelerate deep learning training and large-scale feature modeling tasks. The project will contribute to scalable battery diagnostics methods supporting battery reuse, recycling, and circular energy systems.