Sudipta Patowary

Built a propensity ML model to determine the probability of a customer availing a Personal Loan. The model is able to deliver 91% coverage in the top 3 propensity deciles, thereby identifying an efficient segment of customers to target for marketing the product. Involved in analyzing UPI fraud data to uncover patterns and insights, and build rules to reduce instances of fraudulent transactions. Working on onboarding New-To-Bank customers and modelling their persona and transaction data to predict the best bank products to pitch to them. Promoted within 3 months from Manager Level 1 to Manager Level 2 due to strong performance.

Creating written content for various clients, as well as a short stint creating memes.

Developed a POC Machine Learning application by building and deploying a ML pipeline using Flask to predict factory-issues in PCs; also involved in ETL processes, database management as well as data scraping and Natural Language Processing. Worked on developing a semi-supervised algorithm using Label Propagation as a part of the POC.

Used Ant Colony Optimization and Bayesian learning to build an adaptive learning system which takes data and actions of previous users into account to recommend personalized topics for learning

Worked closely with software development and testing teams in an Agile development environment to design and develop robust C++ and XML solutions to meet client requirements.