Mayank Musaddi

Product Engineer @Sprinklr

Gurugram, IN

Involved in Research, Development and Maintenance of the Entity Linking project, a service that detects and classifies important words in a sentence and then links them to their respective Wikipedia links. This service, written in Python, constantly runs on the whole live Twitter feed and is important in providing live insights to brands about their products. Apart from owning the Entity linking service, I also maintain the Smart Intents service. The smart-intents is a few-shot classification tool that allows businesses to train and deploy a classifier model to detect the intent of the customer message in their respective chatbots. This is the most used service in the Conversational AI suite and my responsibilities involve improvements, benchmarking, deployment and management of these services in different production environments. Involved in the development of a generic infrastructure named Kratos, capable of hosting parallel training and deployment of multiple different machine learning models in a scalable manner. Worked on the research and development of Customer Satisfaction prediction model, a service which rates customer satisfaction based on the conversation flow between the customer and the agent. The work involved data refining, feature selection and experimentation with several state-of-the-art machine learning algorithms.

International Institute of Information Technology

Bachelor of Technology - BTech, Computer Science