Projects

Built a QA based tool to relate every question regarding particular flight manual towards an appropriate paragraphs which contains it’s answer.

Analyzed the various post hoc models used on top of Recommender systems to provide detailed reasoning behind the predictions. A metric based on existing statistical correlation methods is being developed to quantify the degree of it’s explainability.

Collected government documents to build a parallel English-Hindi dataset on sentence level for Neural Machine Translation tasks.

As a part of GSoC project, I have successfully integrated polly loop optimizer of LLVM within Chapel Compiler framework and demonstrated several examples demonstrating the huge speedups gained from these organizations.

With the uneven distribution of privileged features among the users, we are developing semi-supervised models to predict reputation scores for Community QA site users using only commonly available data with the help of teacher model having access to privileged features as well.