Mentor - Prof. Ponnurangam Kumaraguru
My work with Precog lies at the core of my current interests, skills and expertise on social computing systems. I have worked on multiple projects that involve large scale user behavior analysis, studying social interactions and information sharing on social media plarforms. The broad aspects of my work here are:-
Development: I lead the development of a large-scale social media monitoring application, called Advanced Application for Social Media Analytics (AASMA). This effort is funded by the Government of India and is used for monitoring of information spread by drawing inferences from the interactions between users on social networking platforms.
Research: My primary research project here dealt with studying the rising number of selfie-related casuaties happening around the globe, in order to create a technical intervention for the same. The project was in collaboration with researchers from Carnegie Mellon University, and has generated significant awareness worldwide regarding the social issue.
Teaching: During my time here, I have acted as a teaching assistant and a project evalutator for multiple courses. This includes physical courses in IIIT-Delhi like 'Privacy and Security on Online Social Media', and online courses like 'Introduction to Human-Computer Interaction' on NPTEL (India's most popular MOOC portal).
Mentor - Prof. Julien Velcin
My work involved doing a large scale study of the controversy detection schemes on social computing platforms, attempted in the text minning literature. After performing a comprehensive quantitative comparison on a Reddit data set, I developed my own controversy detection mechanism using the comments network on controversial and non-controversial subreddits. Our survey paper on this work is currently underway.
Mentor - Prof. Sanjay Kumar Dhurandher
My work involved application of learning based techniques in Opportunistic Network routing, which hasn't been attempted before. To reduce the high overheads incurred during routing in these networks, we used an unsupervised machine learning technique for routing. This work was a part of my bachelor thesis.
TrulyMadly has earned fame as India’s most popular dating app. TrulyMadly employs moderators to check user profiles for pictures that have NSFW and nude content. I worked as a part of their infrastructure and data science team on building their nudity detection system from scratch. I used the transfer learning framework DeCAF for this work. This will increase the productivity of their moderation process.