* Adidas, Japan: Customer Segmentation * Segmentation to identify distinct customer groups and their characteristic features enabling custom promotional campaigns. * Used clustering algorithms to identify segments of customers. * Panera Bread, USA: Recommendation engine for Catering Customers * Built cross-sell Recommender system in big data environment for catering customers of Panera Bread * Used Item-Item based collaborative filtering as part of Recommender system * Developed a method to assess the quality of recommendation * Land O' Lakes, USA: Sales prediction and Market Mix Modelling * Predicted Dealer sales from shipment data. Market Mix Modelling on sales to calculate ROI from advertisement channels

Part of Data Science team leading machine learning and business analytics initiatives at Cleartrip Role involves designing and implementation of models using prediction, classification and customer segmentation algorithms

Key Projects:-Recommender Engine (Slot Machine Games): Built a recommendation engine to suggest relevant machines to players. Used a combination of Collaborative and content based recommender systemCustomer Segmentation Platform: Built a generic customer segmentation pipeline supporting wide options to preprocess , cluster and profile the dataCheater Identification (Farmville Games) : Built a cheater identifier Pipeline using Isolation trees and helped team to minimize False Positives and maximize the True PositivesLapser Model (Farmville Games): Developed Lapser Model to identify the players who are at the risk of lapsing out from the game. Experimented with multiple Algorithms like Neural Network, Random Forest. Achieved Precision of around 85% with the help of Random Forest Algorithm.Hybrid Recommender System (Draw Something Game) : Developed algorithm to recommend word to draw Show less

* The analysis included finding optimal combinations of vehicle parts to be used together in assembly to minimize the costs while maximizing the performance of system * Used Genetic Algorithm based model to search for the optimal combination of vehicle parts. * Analysis was used as a tool for value engineering in automobile assembly and part selection. * Was part of many new concept vehicle designs as vehicle integrator and data analyst.

*  Built model to predict mileage of a vehicle in real world usage. Used linear regression to build the predictive model. *  Features included various hyper parameters which are inherent to different cities with varying road conditions and population density. *  Handled large dataset to build a holistic model for determining importance of variables impacting vehicle safety