A Machine Learning Practionioner having exposure in all aspects of an ML problem right from brainstorming, building proofs-of-concept to productionizing and maintaining the solution. I am interested in the applications of Machine Learning to solve problems that help make people’s lives easier and augment their productivity by efficient use of technology. I specialize in making NLP solutions for the finance domain and have built a machine learning engine to monitor customer communications by extensively researching and implementing various NLP techniques. Curious and eager to explore more into MLOps and currently in the process of enabling best practices around it at my current workplace. I try to be a better version of myself by being updated with the ongoing research in ML, participating in various reading groups, and actively participating in the ML community.
Zendesk started the customer experience revolution in 2007 by enabling any business around the world to take their customer service online. Today, Zendesk is the champion of great service everywhere for everyone, and powers billions of conversations, connecting more than 100,000 brands with hundreds of millions of customers over telephony, chat, email, messaging, social channels, communities, review sites and help centers. Zendesk products are built with love to be loved. The company was conceived in Copenhagen, Denmark, built and grown in California, taken public in New York City, and today employs more than 4,000 people across the world.
Cognitive View analyzes customer communication to identify Conduct Risk and automates Compliance monitoring process
July 2019 - Nov 2019, Melbourne, Australia
The University of Melbourne is a leading international university with over 160 years experience in teaching and research.
Developed a chrome extension “Get Your Fact Straight” for end-to-end fact checking using Machine Learning(NLP) for Master’s thesis project
Carried out social media analysis to explore the seven deadly sins. The activities were executed on University of Melbourne cloud system - NeCTAR Research Cloud. For social media analysis, Twitter is used as the social media platform, from which tweets are harvested in order to analyse and drive conclusion. The conclusion derived from twitter data on the sins was then cross checked from the official data, made available by AURIN.
For the task of author identification we took into consideration and analysed various machine learning classification models to single out authors for the collected Twitter data set. Alongside the models feature engineering was also applied. Various models were considered, such as Naive Bayes, SVM, Random Forest and neural network multi-class classifiers and how and why they performed under this particular task was reported. In the end, we found that LinearSVC outperformed all the other models by achieving 29% accuracy.
