Aug 2019 – Nov 2021 · 2 yrs 3 mos · Greater New York City Area

Sep 2018 – May 2019 · 8 mos · Charlottesville, Virginia Area

• A research project sponsored by the Army Research Laboratory and the Data Science Institute, UVA
• Tools/Software Used: Python, SQL, Amazon Mechanical Turk, Django Framework, R, Unity, JIRA
• Developed a speech based chatbot within a data-driven framework to train Army personnel to be culturally sensitive
• The system analyzed the cultural competence of a user through speech input via dialogue and decision trees
• Full-fledged product equipped with a VR front end and a backend written in Python Django framework where the speech input is converted into text, classification models are run, feedback selected, and the feedback and score sent via an API
• Leveraged NLP/text mining for feature generation - TF-IDF, LDA, word2vec, and GloVe
• Classification Algorithms used – Logistic, Random Forest, Decision Trees, Gradient Boosting, and Neural Networks. The best model (Gradient Boosting) gave an F1-score of 0.92
• 80% of the users declared that the system accurately judged what they were saying and provided apt feedback
• Paper on Data Generation and Acquisition - https://datascience.virginia.edu/projects/siri-translate-laojia
• Research paper approved for publication in the IEEE SIEDS Journal (June 2019)
• Presented research at the IEEE SIEDS 2019 Symposium, Charlottesville, VA

Jun 2015 – Jun 2018 · 3 yrs · Gurgaon, India

• Key Skills - System Design, Software Development, Product Management & Analytics, Customer Analytics, A/B Testing, Machine Learning
• Developed and executed the functionality of the flight and flight plus hotel vertical of the Goibibo Android mobile app. Spearheaded customer analytics projects pertaining to all business verticals for the mobile application(Android/ IOS). I also implemented several Machine Learning projects that increased the customer lifetime value of the flight vertical
• Personalized offers engine
• Developed the pipeline and the model for personalized offers after segmenting app dropped
• customers by using ML algorithms depending upon the likelihood to book after seeing additional discounts. A retargeting email with the personalized offer would be sent if the booking after the discounts would still be profitable. Of the dropped users, about 8% returned to book the flight with the personalized offer
• Customer Journey Analytics
• Ideated and implemented a user tracking plan for all business verticals from scratch, leveraging various tracking tools to identify funnel drop-off. Led to an overall 4.1 rating in Google Play for the first time in over 1.5 years. Data-driven approach now used for all business decisions, user complaint resolution; led to 10 rich dashboards showing various customer analysis
• Flight grouping using sorting and hashing
• Analyzed search result page (SRP) product impressions data to investigate lower conversion rates in the Flights vertical to optimize the results on the basis of price, baggage options and booking history to increase end-user flexibility. Resulted in a 0.5% increase in click rates from SRP and a 0.3% increase in conversion rates
• Wait-less flight bookings
• Collaborated with the product and design teams to identify user bottlenecks in the Flights booking process to implement a "wait-free" solution that led to a faster booking process, lower app dropouts, a 1% conversion rate increment, and improved user reviews

Jan 2015 – Jun 2015 · 5 mos · Gurgaon

• Key Skills - Process Automation, Data Acquisition and Data Mining
• Developing web scrapers to extract and efficiently aggregate data from various websites (structured/unstructured in format).
• Automated the firm's data parsing and data mining process saving 56 man-months

Jul 2014 – Dec 2014 · 5 mos · Gurgaon, India

• Developed a JSP web application which allowed to track business unit cases and user stories which allowed for better workflow management. Also developed the "export to excel" feature of a generic eclipse GUI plugin and wrote the JUnit test cases for the whole project.