Pay-as-you-go solar home systems are rapidly providing clean and affordable power to the 600 million people in Africa who live beyond the grid. These systems allow people to stop burning kerosene for lighting, to charge their mobile phones, and to have their children study safely at home. However, one core component of the system - the battery - can limit the effectiveness of the system, and the lifetime of the system as a whole. Applying data science and machine learning in an entirely new context, M-KOPA and its research partners will develop new technologies to remotely manage the batteries of its 500,000 customers, and to maximise the lifetime and functionality of these batteries. The research will ultimately lead to reduced product costs, and costs of serving customers, whilst ensuring that solar home system users are able to enjoy the full benefits of clean and safe energy in their homes.