• Spark, Python, R, PostgreSQL, AWS, Tableau, D3, SAS, VBA, Teradata, Hive, C
• Machine Learning, Text Analysis, Time Series Analysis, Predictive Modeling, Data Mining, Image Analysis

Data Science Intern

Williams-Sonoma Inc., San Francisco

• Enhanced accuracy of image classification process from 90% to 99%
• Applied graph cut and mean shift algorithm to identify object in an image and predict its color with 91% accuracy
• Recognized patterns in an image using convolutional neural network with 82% accuracy

Senior Business Analyst

Mu Sigma Business Solutions Pvt. Ltd., Bangalore

Pattern Analysis of Regulatory Visits at Pharmacy Stores

• Predicted likelihood of inspections of stores using logistic regression model
• Performed pre-post analysis, hypothesis testing and k-means clustering
• The model had 75% accuracy and had led to 10% reduction in fines due to non-compliance

Sales Driver Analysis and Sales Forecasting for Spare Parts of Home Appliances

• Developed regression model to determine the various drivers affecting orders of spare parts for distribution channels
• Built forecast model using ARIMA to predict orders of spare parts. Automated process using SAS and VBA

Import Inventory Management for Retail Store

• Forecasted import demand using decomposition time series modeling techniques (SAS)
• Prediction methodology improved demand forecast accuracy by 18%

Compliance Performance Scorecard

• Developed scorecard in Tableau by defining metrics and their thresholds to assign compliance scores to pharmacies

Scalable Yelp Recommendation Service: Built a web app to recommend businesses to Yelp users. Used Spark and applied Matrix Factorization, Random Forest, Gradient Boosting models and feature engineering to predict business ratings. Improved RMS error by 8% over the baseline model of average ratings

Kaggle’s "What’s Cooking" Competition by Yummly: Built multiclass classifier using python, to predict cuisine of a recipe, given a list of ingredients. Built an ensemble of logistic regression, random forest and gradient boosting models and achieved accuracy of 80.99%. Finished amongst top 10% (approx 1400 teams)

REST Web Service: Amazon Beauty Products Data API: Created a RESTful web service (hosted on EC2) using Python Flask to retrieve data, stored in AWS RDS. Performed ETL on json data using python and PostgreSQL

Retweet Analysis: Analyzed factors that lead to retweeting and developed a regression model to predict the number of retweets using Python and PostgreSQL, pulling data from the Twitter REST API

Sentiment Classifier: Classified movie reviews as positive or negative with 82% accuracy, by implementing self-coded Naïve Bayes algorithm in both python and pyspark. Enhanced accuracy up to 90% by adding a subjectivity classifier

SPOT Award
Mu Sigma Business Solutions Pvt. Ltd.

Awarded for ’Pattern Analysis of Regulatory Visits' project for * Leading the project * Providing thorough and rigorously tested analytical solutions * Hard work and dedication