TANMAY GUPTA

• Income Model – XGBoost model to predict income classes and track the change for users based on features extracted from the user’s transactions. Impact: improved accuracy, de-biasing the data, used for specific target marketing. Exposure: Xgboost, Multiclass Classiication. • Aggregate Product - Developed a Financial Data Aggregation model for merchant specific prediction of stock market trends. Added newer dimensions and enhanced evaluation metrics. Built a scalable Spark pipeline to be executed on AWS EMR. Impact: Used by hedge funds, generates revenues in millions of $, Improved model performance, reduced operation cost. Exposure: Big Data, Data Mining & Descriptive Statistics, Statistical Performance evaluation, Code Optimization, User Selection, Outlier Removal, Normalization. • Account Classification Model- Developed an algorithm to identify business accounts based on keywords (fetched using TF-IDF score) and amount filters. Impact: Helped customers to distinguish between personal and SMB accounts. Exposure: Text-processing, TF-IDF Vectorization. • Exploring Various datasets - Analysed possible incoming data sources and was instrumental in providing crucial information about the quality of data and the resulting negotiations. Responsible for maintaining over 3 active projects.

• Intern Project – Built a system that can identify similar merchants from the factual merchant database. Implemented the Word2vec algorithm based on the skip-gram model to find the embeddings of merchant names. Accuracy of predictions was nearly 82%. Exposure: Natural Language Processing, TensorFlow, Word Embeddings.