Built more than 8 predictive models in less than 6 months using SAS enterprise guide and Enterprise Miner. Logistic models were build till variable reduction procedure and then non linear methods are applied later. But sometimes, when the project's timeline is tight, variables selection directly is done from information value(IV) and random forest methods. These models once completed are reviewed(validating on out of time datasets) for deployment process. For variable reduction process, a standard technique is followed. Neural Network, XGboost, Random Forest through H2o, Markov blanket in Bayesian Labs are some of the non linear methods that I have used for the models.R programming is also been used for one of the models and in depth knowledge on R. Large Datasets are pulled from Hadoop and then moved to sas for variable treatment and reduction.