Up-goer summary: “Many people burn wood, dry poop, and dead-animals-that-are-now-rocks to make food and stay warm. I study how doing this makes people sick and how we might help people to stop doing this. Burning these things makes a lot of smoke, and breathing this smoke is very bad. Studies show breathing this smoke kills about four hundred-hundred-hundred people every year. Often, this happens because people just do not have cleaner ways to make food or keep warm. I study ways to understand and fix this. I also study ways to understand how much not-clean air people breathe. Today, finding out exactly how much not-clean air people breathe takes a lot of money. This is because computers that read smoke numbers very well are a lot of money. But some computers that read smoke are less money to own because they are not very good at it. I want to turn these bad smoke reading computers into good smoke reading computers that are not much money to own by giving them a brain.” Drew Hill studies the impacts of energy use on the environment, human health, and climate. Most recently, his research has focused on characterizing the link between household solid fuel use and health in a way that informs energy, health, and economic policies. More than 3 billion of the world’s poorest people rely on wood, coal, dung, or crop residues to meet their daily energy needs. These fuels, which are most often employed in open fires or poorly ventilated stoves, billow dirty smoke into users’ kitchens and into the local air. The pollution exposures that result are estimated to lead to over 4 million deaths annually, considerable economic loss, stifled social development, and rapid local climate change. Good Green Ideas Recommendation: I am increasingly interested in the effects of exposure assessment study design on health and energy policies, specifically the misclassification of exposures to PM2.5. The SAGE IGERT program has allowed me the flexibility to reach beyond my primary PhD designation and explore the utility of crowd sourced air quality monitoring networks in reducing such misclassification. I am currently working with colleagues at the School of Public Health to develop a hyper-local network in the East Bay for the purposes of examining the role of machine learning techniques in calibrating low-cost sensor networks.