Kumar Kovid

• Engineered a model to predict best target population for Insurance Agents, leveraging health, demographic, and habitual data, leading to 25% improvement in conversion rate.
• Automated commission calculation by extracting data from XML’s and applying multi-stepped business logic to remove inconsistencies.
• Programmed a real-time lineage graph of data flow in the Data Management system, using Python & SQL, providing an enterprise level status of all the ETL jobs running at any point in time.
• Designed an HQL, User Defined Functions for Hive, Bash Script & Python based solution to resolve the conflict between Prospect & Customer Databases, projected to generate about $4m additional revenue over the next 3 years.
• Collaborated with a team of 2 to create customer dashboard with lifetime value, attrition probable & best target prospects.

• Discovered 200 additional Money Laundering cases worth millions of dollars, from billions of transactions that occur within Deutsche Bank eco-system, by improving the models written in Python and Oracle SQL.
• Led a team of 3 people to write models to detect additional suspicious financial activity, adding 2500 additional transactions to suspicion list.
• Saved 12 work hours/week by automating ad-hoc reporting by utilizing VBA, VLOOKUP, pivot tables, etc.

• Spearheaded 10% improvement in on-time project deliverables by acting as a SME liaison between IT & business teams.
• Modeled most likely to churn & reasons of high churn for employees, leveraging SVM, Random Forest & correlation models.
• Implemented efficient SQL queries saving 20% time to develop mission-critical insights & presented findings to executives.
• Optimized & managed Informatica workflows successfully & generated pivot reporting using advanced Excel & Python.

• Identified nearly 4K previously unknown property tax evasion cases across all 5 boroughs of NYC by combining & analyzing trends & outliers in property, fire, water & permits data using NumPy & Pandas in Python & SAS.
• Investigated 20 million combined rows from multiple public datasets to discover 1K new illegal constructions in NYC, utilizing regression, classification & statistical framework models in Python.
• Modernized existing methods for data analysis, & persuaded senior management to adopt new & updated methodology.

• Devised & maintained ETL mappings to extract data from multiple sources to the Staging area, EDW & then to Data Marts.
• Handled large datasets using Spark in Memory capabilities, using broadcasts variables in Spark, effective & efficient joins, transformations, & other capabilities, to save 10 working hours per week.

GPA: 3.945/4