Sutirtha Bhattacharyya

Analyzed large volumes of signal data from mobile devices through PySpark to uncover travel patterns and establish traffic trends in any given area which was added to the product suite of the company, resulting in a 10% increase in revenue. Designed and implemented a data pipeline that scrapes and processes the data of doctors from various online sources, and an ML model that uses said data to predict the relative importance of the doctor with up to 85% accuracy. Processed and analyzed ATM transaction data of an ATM vendor company and used it to create a forecasting model to predict the future transaction volume of their ATMs with 75% accuracy, resulting in an increase in revenue for the vendor. Created an optimization algorithm for a retail chain to find the optimal delivery person to carry out a delivery and also the optimal route they must take to perform the delivery, which reduced their average delivery time by 30%.

Developed a system to perform automated merchant underwriting by training an ML model on customer data to predict the reliability of the customer with up to 90% accuracy, and then deploying the model as a service to monitor its performance metrics. Designed an ML pipeline to extract and process transaction data of a fintech company, train a ML model to predict fraud transactions up to 89% accuracy, and deploy model on as server via an API to be used by the client. Developed a deep learning-based facial recognition application using TensorFlow that identifies the customers of a bank from a video feed and automatically fetches their KYC details, which improved the operational efficiency of the bank. Developed an AI-powered chat bot using Google Dialogflow which answers the queries of the customers of a fintech firm and also performs transactions as per their suggestions, which completely automated their customer management system.

Specialization in Machine Learning.