United States

Data Scientist

Master's degree

Bachelor's degree

Roshan Kumar

Staff Data Scientist

Analytical skills

Advanced data analytics

Python

Databricks

•	Led a team of data scientist and data engineers to build data products. Helping the business with 
        strategic decisions. Promoting a data driven culture. 
•	Forecasting, Anomoly detection, Clustering of accounts
•	Build and deploy ML models at scale adhering to MLOp best practices
•	Nominated for Microsoft AHSI Impact Award
•	Received Microsoft Excellence Award for building analytics platform for Impact Dashboard

• Churn prediction : Built and deploy end to end churn prediction model in AWS Sagemaker using RF
• Collaborated with cross-functional teams to understand their business needs, and formulate and 
   complete end-to-end project lifecycle that includes data gathering, analysis, and presentations.
• Text mining : Analyzed the call center chat scripts to predict the reason for customer inbound calls 
   using tf-idf

- Won Allstate Business Challenge to work with Allstate on year-long analytics project. Awarded cash prize for $30,000
-Built a recommendation system for personalized rewards for customers for Allstate drivewise app

- Forecasted the demand for UK sales (headphone category) with 95% accuracy using ARIMA time  
   series model which prevented overstocking of products
- Received the highest award at an organization level “Individual Award for Innovation” in recognition of
   influential research work and ensuring timely deliverables

- Built online purchase intent prediction model for telecom operator Optus using Naïve Bayes classifier
- Improved the model by introducing the contextual invite feature for different customer segment using
  decision tree which led to 10% increase in response rate among customers

- Analyze the dynamics of current customers to determine future trends in customer activities.
- Implemented various modelling techniques (logistic regression, time series, market basket) in R
- Developed an application- Retail Customer Analytics(RCA) used to predict attrition, cross sell, 
   product propensity, correlated products in a basket.

Implemented services using Service Factory function of OMISCID library and communicate between different services using connectors.

• Estimated the motion of an object from a sequence of images extracted from the video of a moving
   entity.Tracked object features and correspondence between images using KLT (Kanade-Lucas- 
   Tomasi) algorithm implemented in C++

Roshan Kumar🇺🇸

About Roshan Kumar

Expertise

Sessions with Roshan Kumar