Pranav Dev Kumar

Developed libraries and introduced processes to improve existing automation process, that would save company over 100 hours of development time. Led the analytics of a new department by creating reports, roadmaps, plans and process structures that improved work efficiency by 50%. Led the development of lead quality model for sales that improved efficiency of sales personnel by 20% over previous iteration of the model.

Devised and deployed lightgbm models that increased customer approvals by 20% and reduced default rate by 29%. Built a hyperparameter optimization library, resulting in 20% improvement in model results and reduction in training time by 15%. Led a team for the development of automated ML solutions and organised roadshows with clients and helped them onboard onto the new product. Initiated a custom of gifting books on personal finance management to new hires to improve morale and financial health.

Created machine learning models using algorithms like logistic regression, XGBoost and Light-Gbm to create ensemble learning models that reduced client risk by 10% and increased approval rates by 20% on their live data. Utilized alternate sources of data like device data, SMS data, location data etc for predicting risk across various client portfolios. Optimized 8 data pipelines to reduce latency from 24 seconds to 4 seconds. Worked on creating structured data from unstructured SMS data using regular expressions. Developed polished visualizations to share results of data analyses using Grafana which were able to highlight serious latency issues with FinBox’s premier product.

Risk analysis and strategy alignment for new product by company. Coordinated launch of new product on which over 500 new customers were onboarded. Reduced risk of default by over 30% for their premier product

End to end process ownership of incentive compensation process for over 100 sales professionals. Responsible for maintaining client SLA for real time support and reporting. Goal setting and tracking progress of various sales representatives for US based client. Created Natural Disaster impact analysis that corrected sales incentives for over 50 sales professionals conducting sales in specific disaster struck areas.