Manager Data

Piyush Sharma ; Linkedin

**** ***** ******, ******* ** Cell:402-***-**** ; e-mail: ****************@*****.***

Summary and Skills

An astute professional with 10 years of experience of using data science and analytics for marketing, consumer behavior, pattern/trend analysis, customer segmentation, pricing and demand forecasting.

Experience of handling multiple clients as well as multiple projects at a time.

Experience of mentoring team and working closely with senior executive team in finding new opportunity to leverage data science and reporting financial impact of predictive models.

Hands on experience of implementing deep learning, supervised learning (Regression, classification, Neural Net) and unsupervised learning (clustering) from scratch with advance algorithm like Bayesian modeling, Random Forest, stack modeling etc.

Tool: Tensor flow,Apache Spark,Hive,Python, Scala,SQl, Tableau, R,, Advance Excel

Six Sigma Green Belt certified.

Professional Experience

Data Scientist Manager (Elevate Credit) June 2017 – Present

Leading a team of four data scientist to provide underwriting method for loan applications

Developed a deep learning model for fraud detection resulted in 10% reduction in first party fraud.

Led development of internal data by combining data from several bureaus which resulted in 7 % reduction in charge-off revenue.

Led APR and fond amount optimization project to convert 30% applications to loans resulted in 10 Millions USD of additional revenue per month.

Develop attributes from bank transaction data using text mining and develop a model to predict fraud

Data Scientist Manager (Nebraska Book Company) November 2012 – May 2017

Working with CEO office to provide quantitative insight for strategic growth of company.

Leading data science team in Nebraska Book Company that provide analytic insight for strategy, product development, inventory planning, pricing and reporting.

Successfully developed three products from concept to sales support and pricing, resulted in 10% growth in revenue.

Built the pricing strategy model for ecommerce market place and commercial clients to optimize revenue and gain market, resulted in more than $ 2 million gain in revenue using Python and Bayesian Modeling.

Developed a model for Email targeting for marketing ads, resulted in 23 % lift in response using Python and XGboost.

Developed a dashboard to measure KPI using Tableau and SQL.

Developed demand planning time series model resulted in more than $ 1 million saving in inventory using ARIMA.

Customer segmentation to develop customer specific marketing strategy by cluster analysis.

Developed model to determine life cycle of the book using machine learning using stack model in Python.

Data Mining Manager (Alorica Inc.) May 2015 – Sept 2015

Leading a team of three analysts to develop statistical model used for optimization of debt collection.

Build customer win back model for one of the biggest cable company resulted in more than 20% lift in efforts using Python, SQL and Logistic Regression.

Developed a model to predict the length of call and factors contributing repeat call resulted in 15% reduction in call duration and 30 % reduction in repeat rate using ensembles and stack model.

Research Analyst (University of Missouri Hospital) August 2011 – July 2012

Developed Decision Analysis System using process mapping, usability, and data mining to improve efficiency.

Develop a model to determine the length of stay and resource allocation for the patient at registration.

Senior Engineer (Product Development) (Samsung Electronics) March 2006 – December 2009

Worked with marketing team to develop models which address customer problem and reduce market defect 50%

Optimized call handling process by call segmentation using data mining methods resulted in INR 40 Lakhs service cost and increased customer satisfaction.

Developed a model to determine most desired features for refrigerators, won best model of year award.

Academic Credentials

M.S., Industrial and Manufacturing Systems Engineering, University of Missouri, Columbia, USA GPA: 3.33/4.0 (June 2012)

B.S., Mechanical Engineering, National Institute of Technology Durgapur, INDIA

Professional Projects

1.Internal Bureau Data development in Elevate

Since all the lenders doesn’t report to all credit bureaus for single person different bureaus can report different inquiry and trade data. Scope of this project was to combine the raw data from various bureaus and create elevate custom attributes. These Elevate custom attributes provided lift of 7% additional revenue and 15 % reduction in charge off losses.

2.Optimizing APR and loan fund amount

30% of applications become non-funded because of high APR and the amount offered to customers. Developed a simulation and optimization model to predict the probability of customer accepting the loan. Resulted in 20% increment in acceptance rate

3.Revenue optimization for E-commerce (Nebraska book company)

Provided 50% growth for ecommerce by pricing and inventory optimization.

Tool used Python, SQL and linear optimization and probabilistic clustering.

4.Demand forecasting software( Nebraska Book Company)

Developed a forecasting algorithm using stack modeling using python and SQL

Managed project from idea to final implementation

Overall inventory management was improved by 30%

300 stores signup for this product

5.Account and inventory segmentation (Nebraska Book Company)

Developed a model to segment books and account using Python and SQL

Gained 15 % reduction in return rate and 22 % improvement in acquisition

6.Improving debt collect( Alorica)

Client: Hospitals

Led a team of three data scientist to developed a model to segment agents in to different deciles based on their propensity to pay

Achieved 10 % reduction in operational cost and 5% increment in debt collection

It was two level models, on level 1 we used logistic regression, XG boost and Random forest and level 1 model was developed using the results of all level 0 models

7.Customer win back model (Alorica)

Client: Direct TV

Developed a model to classify lost accounts who are more likely to rejoin direct TV

.Imbalanced class classification model, developed using python.

Roc curve was used to evaluate performance of the model

8.Customer risk and fraud model for credit union (Alorica)

Client: Credit Union

Developed a model to identify customer who are more likely to borrow money

Risk analysis was performed using credit data

Python and stacking algorithm were used to classification

9.Market analysis and customer segmentation (Samsung)

Design and analyze survey data to understand customers need for new model

Features were classified based on the propensity to pay extra money by customers

The product developed after analysis won best product of the year.

Contact this candidate