Sandeep Rajput EMAIL: sandeep_rajput@hotmail.com PHONE: (425) 449 9554 [Cell] WEB: www.sandeeprajput.com Professional Summary 13 years in Predictive/scientific analytics development and management in Industry; 5 in academia 5 years team management 4-8 FTEs and 20-30 contractors distributed globally; 8 years as team lead Knowledgeable about classical and modern tools and technologies (C++, SAS, SQL to Java, R, Hadoop) Professional Experience 2010—2013 PRINCIPAL DATA SCIENTIST & GROUP PRODUCT MANAGER (MONETIZATION), Microsoft Built models and algorithms to flag harvested or pirated Win7 license keys worldwide from large data. Modeled and identified signatures of piracy for several licensing channels Led keystone projects to unlock advertiser demand on adCenter (aka Bing Ads) by 40-60% through specific bid, keyword and campaign guidance at scale through scientific analysis of massive data DATA SCIENCE DIGITAL ADVERTISING Created demand and supply slices via data mining on Bing Ads monetization data; modeled revenue displacement and other metrics for effective use of marketing resources BIG DATA 2003—2010 SENIOR RESEARCH SCIENTIST LEAD, Fair Isaac Corporation (FICO) PREDICTIVE MODELING Built suites of models (regression, neural networks, additive models) to predict delinquency, likelihood to pay, revenue, profit, attrition, payment card and e-commerce fraud for clients in North America, Europe and Middle East; advised on optimal strategies for these outcomes MACHINE LEARNING Developed feature libraries of consumer preferences from time-series and textual information, predictive models with these features helped generate new line of revenue PROD. DEVELOPMENT Applied R&D evaluating alternate predictive algorithms, feature selection methodologies and using multiple scores to optimize customer lifetime value; implemented model monitoring schemes Designed the 2009 FICO/UCSD Data Mining Contest (304 teams from 35 countries contested) DATA MINING Co-founded the internal wiki on predictive modeling for training and reference; served as subject matter expert on predictive modeling process, from data summary to post-deployment activities RETAIL BUSINESS Consulted to optimize store location and targeting; analyzed campaign performance; built direct marketing response/conversion models; helped optimize product offerings 1996—1998 ASSISTANT MANAGER (PROJECTS), Reliance Industries Limited TECH. PROGRAM Managed the revamp of a Polyester Staple Fiber (PSF) plant to double production; created equipment specs, supervised technical output of vendor engineers; designed the DCS dashboard MODELING/SIMULATION Technical consulting for executive engineers; created math models to fit observed measurements, simulated control strategies to help reduce downtime and accidental shutdown 1998—2003 GRAD. RESEARCH ASSISTANT, Measurement and Control Engineering Center SCIENTIFIC MODELING Developed a Matlab toolbox to track small particles in fluidized beds from videos/movies to characterize the velocity fields; compared experimental output with CFD simulations DATA SCIENCE Consulted with large chemical companies for fault diagnosis and monitoring via nonlinear time-series analysis; Developed GUI software package for nonlinear time-series analysis in Matlab; held workshops for industry attendees, created training material and manuals Education Ph.D. CHEMICAL ENGINEERING The University of Tennessee, Knoxville 1998—2003 M.S. STATISTICS The University of Tennessee, Knoxville 2002—2003 B. S. CHEMICAL ENGINEERING Indian Institute of Technology, Kanpur, India 1992—1996 Technical Skills PROGRAMMING ANALYSIS BIG DATA MODELING MACHINE LEARNING C++, Java, Perl, Python, Scala, Fortran; Eclipse, NetBeans, Visual Studio; Subversion, Maven Matlab, R, SAS, NumPy/SciPy/Pandas, Octave; LaTeX, Beamer; MS Word, Excel, PowerPoint, Visio Apache Hadoop, Hive, Pig; Mallet, OpenNLP, Lucene/Solr; SQL Server, PostGreSQL Linear, logistic and generalized regression; Generalized additive models; Bayesian models Neural networks, Clustering; Mixture Models; Random forests; Decision trees Applied Research 2014 Web users as Automatons with limited Sentience: a Physics-based model of user interaction. 2013 [w/Paul Smolikov] Segmenting Web users by informational needs: A case study on Bing users. Heavy tails in Online Experiments: Power laws and preferential attachment. Measuring scale in second-price auctions. 2012 Six tropes and the alignment of demand and supply. 2011 Search user intent and the primacy of local time. 2010 Modeling Search marketplace metrics with Robust AR models. 2009 Neural Networks and Special Values: Building better predictive models. 2008 Next Best Action: a Reinforcement Learning paradigm. 2007 Recursive profiling and its impact on model performance. 2005 Event-triggered marketing: reaching customers at the right time. 2004 Dynamic marketing strategy to detect customer lifestyle changes. Academic Research 2004 2003 Sarnobat, S. U., Rajput, S., Bruns, D. D., DePaoli, D. W., Daw, C. S. and Nguyen, K. (2004). Impact of external electrostatic fields on gas-liquid bubbling dynamics. Chem. Eng. Sci. 59(1), 247—258. Rajput, S. and Bruns, D. D. (2003). Nonlinear time series analysis of flooding in a distillation column, Paper 465d, in Proc. AIChE Annual Meeting 2003. ISBN 0-8169-0941-5 Rajput, S. and Bruns, D. D. (2003), Principal Curves and Chaos, AIP Conf. Proc., 676(1), 327-332. Rajput, S. and Bruns, D. D. (2003). Numerical simulations of the fluidized bed experiments using MFIX multiphase CFD code. Report for Oak Ridge National Laboratory, ORNL-400002312 Rajput, S. and Bruns, D. D. (2003). Detection of Velocity Fields from videos of particles in fluidized beds. Report for Oak Ridge National Laboratory, ORNL-400002312. 2002 Rajput, S. and Bozdogan, H. (2002). Choosing the number of PCs in localized PCA using kernel smoothing and information-theoretic criteria. Report for Statistics Dept., UT Knoxville. 2000 Rajput, S., Shul-Cloper, R., Abidi, M. A. and Gonzalez, R. C. (2000). A new method for searching an image in a scene. Report for IRIS Lab, Elec. & Comp. Eng. Dept., UT Knoxville.