A Unified Customer Insights Platform for FinTechs.

• Responsible for architecture design and development of a new persona named Sally, to handle first notice of loss during insurance claim filing. Based on intent classifiers using trained language models, word embeddings on audio transcripted data, domain level reports and finite statemachines for conversation flow.• Developed and deployed deep learning based image classification models for roof shape and pool identification from aerial imagery which helped the company introduce 2 new columns in the Property Database for decision making.• Working on to build a recommendation engine for claims line items during the claim inspection process. Based on an ensemble of the trained sequence to sequence models and word2vec on historically inspected claim reports.• Contributed to a key driver analysis module to analyze features contributing to the risk of properties in homeowners insurance. Based on the random forest, gradient boosting tree-based classification models.

• Contributed and maintained a product which helped corporates and PR professionals in risk and reputation management through social mediaanalysis and provides a predictive analysis of concerned issues going over the globe.• Developed trained classifier to tag tweets into different issue categories. Based on pre-trained word embeddings.• Implemented a scoring module to generate issue risk scores by analyzing and aggregating tweets’ metadata over a time window.• Contributed to trend prediction module to calculate growth/decay factor calculated by aggregating the scores during a given time frame

• Developed back end for the user, session management and mailing service module for child organization edu4sure.com. Based on python flask framework.• Implemented back end for CRUD services for application admin panel. Based on the flask.