I am primarily interested in scalable inference, i.e. drawing conclusions from large amounts of inconsistent data. This problem is a key computational bottleneck for modern applications such as information extraction, knowledge aggregation, question-answering systems, computer vision, and machine intelligence. Please read one of the following articles to get an idea of my research: (approximate lifted inference, bounding Boolean functions, linearized belief propagation). And here is a recent CACM 2015 article by Stuart Russel that justifies why we should aim to combine methods from databasess (first-order logic plus scalability) with those from machine learning (statistical reasoning plus learning).