Oct 2020 – Nov 2022 · 2 yrs 1 mo

• Responsibilities: -
• Founding Member of the Tech Team of Zuperly that built the entire SAAS product from scratch
• Architected and worked on the Tech stack of Next.js, TypeScript, Node.js, and Ruby on Rails
• Key Projects: -
• Zuperly Web Clipper Chrome Extension: -
• Sole owner of the Chrome Extension Development, Deployment, and Monitoring
• One-click resource capture feature of Extension increased the key metric of resources captured on the dashboard
• Built Real-time Nudges and Notifications Engine which increased daily and weekly active users
• Zuperly Web Application
• Member of a 3-person team that developed Front-End React Components as required for the Web Application
• Lead a 2-person team that built the Real-time Collaborative Editor based on Tiptap Editor
• Added socket.io based real-time data sync among Backend, Web app, and Zuperly Extension for showing up-to-date data resulting in improved experience for the users

Jan 2020 – May 2020 · 4 mos

• Worked as a software intern for backend development in Django and using NLP and text analytics to derive insight into the data.

Aug 2018 – Jul 2019 · 11 mos · Gainesville Metropolitan Area

• Part of the NLP team, researching dialogue systems.
• Researching Natural Language Generation using Deep Learning Models and ways to improve them in various tasks such as story generation and task-oriented chat-bots.

Feb 2018 – Jun 2018 · 4 mos · Pune, Maharashtra, India

• Projects Undertaken:
• Ticket Resolution Chatbot
• Designed and developed a cross-platform mobile-based interactive ‘Employee Ticket Submission’ chatbot.
• Increased intuitiveness and efficiency of the current ticket system resulting in a 50% increase in system usage.
• Implemented the ‘wait time predictor’ module in TensorFlow to achieve 93% accuracy in estimation of issue resolution time within a day’s timeframe.
• Smart Real Estate Recommendation System
• Developed a recommendation algorithm for an Australian real estate client using regression models and NLP.
• Conceptualized, developed and maintained the code for a separate recommender system for real estate agents to enable them to advise the potential customers, resulting in an overall increase of 20% in the user base. Improved recommendation accuracy from 65% to 85% while obtaining ~96% among immediate buyers.
• Decentralized EMR Healthcare Application
• Contributed to building a Block-chain-based Full-stack Decentralized EMR Healthcare application. The application was developed using the Hyperledger Composer framework utilizing Docker and Node.js framework.

Jul 2016 – May 2017 · 10 mos · Pune, Maharashtra, India

• Researched and worked on the idea of 'Resequencing of news articles sentimental analyses' as a final year undergrad academic project, in a four-member team
• This was the part of the research about 'AI-powered newsfeed representation' best on creating linkages in different types of interrelated news articles and representing them in such a way that overall positive sentiment value of the news feed should be maximum.
• My responsibility was to perform data-preprocessing, clustering and topic extraction from news text. Sentiment analysis was done using Stanford's CoreNLP module. Latent Dirichlet Allocation was used for topic modelling. Hierarchical clustering was used to group and cluster stories together.
• Tools and frameworks used: Python3, NLTK, Gensim, scikit-learn, CoreNLP