-Developed ranking algorithms for the purpose of sorting items in response to user queries on the live website (primarily gradient boosted machines and ranking SVMs)- A/B testing and monitoring of model variants, including both offline and online statistical analysis and validation.- Developed classification and lookalike algorithms for the optimization of marketing spend.- Development of time-series models to predict the behavior of certain markets and the price of various commodities.

Data AnalysisDuring my time on the CMS experiment I worked on both Higgs and supersymmetry based analyses. I was the co-convenor of the WH2HadTau Higgs subgroup, focusing on searches for the associated production of the Standard Model Higgs boson and a W± boson, where the Higgs boson decays to a pair of tau-leptons. These searches employed a variety of multivariate techniques to attempt to extract possible signals from data, including boosted decision trees, neural networks and k-nearest neighbour approaches.Technical Work– Tau triggerAs part of my technical commitment to the CMS experiment I developed a new trigger to select events containing a pair of tau-leptons. This trigger was significantly improved with respect to the previous iteration, and employed a novel region of interest selection in which to perform particle tracking extremely quickly. – Muon alignmentAs a group at Texas A&M we developed an algorithm that performs a precise track-based alignment of the CMS muon system. A detailed understanding of the alignment of the muon system is crucial for the reconstruction performance of high momenta muons. The track-based algorithm is based on the Map-Reduce programming model, and optimises the position of the chambers with respect to the silicon tracker via a multidimensional fit, using muon tracks traversing the chambers.