► Developed a framework for anomaly detection for arbitrary metrics using deep learning based forecasting methods► Developed machine learning models targeting several kinds of fraud in telecom space along with an end to end high-performance pipeline for real-time feature calculation ► Designed and implemented framework to identify deviant behavior in customers through root cause analysis for a large insurer► Delivered customer micro segments and customer journey attribution model using Combined Influence trees and Bayesian Networks for a large fashion e retailer with over $2 bn of revenue► Developed customer churn models(Logistic Regression) for a large insurer helping them achieve retention rate of 80%► Involved clients in the analysis journey with info graphics and presentations to help them understand findings and insights► Designed and implemented a reinforcement learning system based around mimicking parts of the brain (Hierarchical Temporal Memory) for on-line time series prediction reaching 95% accuracy

•Responsibilities:Client side interaction and delivery management, primarily involved in problem identification, data understanding and extraction, data cleaning and statistical modelling.Projects:-1)Finding recurring patterns in fraudulent claims-Creating models to identify frauds perpetrated by medical facilities and providers and identify the segment of providers likely to commit fraud.2)Log data:-Devised a deterministic model for identifying the IP address through which brute force attack done.3.Scoring Patients and Facilities for a Health Care Service Provider :-a)Devised a statistical measure for patients’ “suffering” using CAHPS survey responses as one of the inputs.b)Created an alerts mechanism to identify low performing facilities on the patient suffering spectrum.c)Received special commendation for my performance in the project.4. Benchmarking Sales Data :-a)Developed a statistical measure DeaDRIX (Dealer Distributor Relative Index) for relative ranking of Dealers with respective Distributor based on frequency and quantity of sales

1. Using computer vision/ image processing/ deep learning for various applications2. Processing sensor data of accelerometer, gyroscope, magnetometer, and GPS mounted on a vehicle for determining road conditions, vehicle speed, acceleration, braking, and other activities3. Working on product development of an Intelligent Electric Scooter with smart features where machine learning and data analytics have huge applications such as increasing safety of the user, road profiling, rider profiling, predictive maintenance, sensor data processing, optimizing rider experience, and vehicle performance