Oct 2012 – May 2014 · 1 yr 7 mos · Redmond

May 2011 – Jul 2011 · 2 mos · Hyderabad Area, India

• The project aim was to develop a prototype for mining feedback data submitted by the users of a software product with prominent feature being automatic classification of incoming comments into predefined categories like Good/OK/Bad and into Feedback/Functional/Others category. The classification accuracy was nearly 87% using PLSA algorithm. I also implemented EM Algorithm for boosting accuracy and Shannon and Renyi Entropy as error detecting measures.

Aug 2010 – May 2012 · 1 yr 9 mos

• Founded Software Development Section Labs (SDSLabs) of IIT Roorkee to promote programming & software projects in the college.

May 2010 – Jul 2010 · 2 mos · Minneapolis, Minnesota

• I worked under Dr. Vipin Kumar, HOD Computer Science and Electrical Engineering Department, University of Minnesota as an intern. I helped in developing optimized segmentation based mining algorithms on data streams for detecting land cover changes with high degree of accuracy based on analysis of EVI and FPAR time series of the locations. The aim was to flag the locations as change points which have undergone major land cover changes in the past ten years. This problem is of considerable importance for climate monitoring groups as it provides information about precise locations of deforestation, floods, etc. Since this approach is based on time series analysis, it gives superior performance in comparison to image processing based approaches and is suited for further development in order to tune it to real time analysis.